Publish binary from main instead of discarding

This removes invalid states and simplifies pattern matching.
It also makes it easier to add further address extensions.

.github/workflows/GnuComment.yml

@@ -0,0 +1,80 @@
+name: GnuComment
+
+on:
+  workflow_run:
+    workflows: ["GnuTests"]
+    types:
+      - completed
+
+permissions: {}
+jobs:
+  post-comment:
+    permissions:
+      actions: read  #  to list workflow runs artifacts
+      pull-requests: write  #  to comment on pr
+
+    runs-on: ubuntu-latest
+    if: >
+      github.event.workflow_run.event == 'pull_request'
+    steps:
+      - name: 'Download artifact'
+        uses: actions/github-script@v8
+        with:
+          script: |
+            // List all artifacts from GnuTests
+            var artifacts = await github.rest.actions.listWorkflowRunArtifacts({
+              owner: context.repo.owner,
+              repo: context.repo.repo,
+              run_id: ${{ github.event.workflow_run.id }},
+            });
+
+            // Download the "comment" artifact, which contains a PR number (NR) and result.txt
+            var matchArtifact = artifacts.data.artifacts.filter((artifact) => {
+              return artifact.name == "comment"
+            })[0];
+            
+            if (!matchArtifact) {
+              console.log('No comment artifact found');
+              return;
+            }
+            
+            var download = await github.rest.actions.downloadArtifact({
+              owner: context.repo.owner,
+              repo: context.repo.repo,
+              artifact_id: matchArtifact.id,
+              archive_format: 'zip',
+            });
+            var fs = require('fs');
+            fs.writeFileSync('${{ github.workspace }}/comment.zip', Buffer.from(download.data));
+      - run: unzip comment.zip || echo "Failed to unzip comment artifact"
+
+      - name: 'Comment on PR'
+        uses: actions/github-script@v8
+        with:
+          github-token: ${{ secrets.GITHUB_TOKEN }}
+          script: |
+            var fs = require('fs');
+            
+            // Check if files exist
+            if (!fs.existsSync('./NR')) {
+              console.log('No NR file found, skipping comment');
+              return;
+            }
+            if (!fs.existsSync('./result.txt')) {
+              console.log('No result.txt file found, skipping comment');
+              return;
+            }
+            
+            var issue_number = Number(fs.readFileSync('./NR'));
+            var content = fs.readFileSync('./result.txt');
+            
+            if (content.toString().trim().length > 7) { // 7 because we have backquote + \n
+              await github.rest.issues.createComment({
+                owner: context.repo.owner,
+                repo: context.repo.repo,
+                issue_number: issue_number,
+                body: 'GNU sed testsuite comparison:\n```\n' + content + '```'
+              });
+            } else {
+              console.log('Comment content too short, skipping');
+            }

.github/workflows/GnuTests.yml

@@ -0,0 +1,284 @@
+name: GnuTests
+
+# Run GNU sed testsuite against the Rust sed implementation
+# This workflow extracts and runs tests from the GNU sed testsuite to ensure compatibility
+
+on:
+  pull_request:
+  push:
+    branches:
+      - '*'
+
+permissions:
+  contents: write # Publish sed instead of discarding
+
+# End the current execution if there is a new changeset in the PR
+concurrency:
+  group: ${{ github.workflow }}-${{ github.ref }}
+  cancel-in-progress: ${{ github.ref != 'refs/heads/main' }}
+
+env:
+  DEFAULT_BRANCH: ${{ github.event.repository.default_branch }}
+  TEST_FULL_SUMMARY_FILE: 'sed-gnu-full-result.json'
+
+jobs:
+  native:
+    name: Run GNU sed testsuite
+    runs-on: ubuntu-24.04
+    steps:
+    #### Get the code, setup cache
+    - name: Checkout code (sed)
+      uses: actions/checkout@v6
+      with:
+        path: 'sed'
+        persist-credentials: false
+    - uses: dtolnay/rust-toolchain@master
+      with:
+        toolchain: stable
+        components: rustfmt
+    - uses: Swatinem/rust-cache@v2
+      with:
+        workspaces: "./sed -> target"
+    - name: Checkout code (GNU sed testsuite)
+      uses: actions/checkout@v6
+      with:
+        repository: 'mirror/sed'
+        path: 'gnu.sed'
+        ref: 'master'
+        persist-credentials: false
+
+    # Omit installing part of https://github.com/actions/runner-images/tree/main/images/ubuntu
+
+    ### Build
+    - name: Build Rust sed binary
+      shell: bash
+      run: |
+        ## Build Rust sed binary
+        cd 'sed'
+        cargo build --config=profile.release.strip=true --profile=release #-fast
+        zstd -19 target/release/sed -o ../sed-x86_64-unknown-linux-gnu.zst
+    - name: Publish latest commit
+      uses: softprops/action-gh-release@v2
+      if: github.event_name == 'push' && github.ref == 'refs/heads/main'
+      with:
+        tag_name: latest-commit
+        draft: false
+        prerelease: true
+        files: |
+          sed-x86_64-unknown-linux-gnu.zst
+      env:
+        GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+
+    ### Run tests
+    - name: Run GNU sed testsuite
+      shell: bash
+      run: |
+        ## Run GNU sed testsuite using our script
+        cd 'sed'
+        # Set GNU testsuite directory
+        export GNU_TESTSUITE_DIR="../gnu.sed/testsuite"
+        # Run tests with JSON output
+        ./util/run-gnu-testsuite.sh --json-output "${{ env.TEST_FULL_SUMMARY_FILE }}" || true
+
+    ### Upload artifacts
+    - name: Check for JSON results file
+      shell: bash
+      run: |
+        echo "Checking for JSON results file..."
+        ls -la sed/ || true
+        ls -la sed/${{ env.TEST_FULL_SUMMARY_FILE }} || echo "JSON file not found at sed/${{ env.TEST_FULL_SUMMARY_FILE }}"
+        ls -la ${{ env.TEST_FULL_SUMMARY_FILE }} || echo "JSON file not found at ${{ env.TEST_FULL_SUMMARY_FILE }}"
+        find . -name "*json*" -type f || echo "No JSON files found"
+
+    - name: Upload full json results
+      uses: actions/upload-artifact@v6
+      with:
+        name: sed-gnu-full-result
+        path: sed/${{ env.TEST_FULL_SUMMARY_FILE }}
+        if-no-files-found: warn
+
+    - name: Upload test logs
+      if: always()
+      uses: actions/upload-artifact@v6
+      with:
+        name: test-logs
+        path: |
+          sed/test-logs/*.log
+          sed/test-results/*.json
+
+  aggregate:
+    needs: [native]
+    permissions:
+      actions: read
+      contents: read
+      pull-requests: read
+    name: Aggregate GNU test results
+    runs-on: ubuntu-24.04
+    steps:
+    - name: Initialize workflow variables
+      id: vars
+      shell: bash
+      run: |
+        ## VARs setup
+        outputs() { step_id="${{ github.action }}"; for var in "$@" ; do echo steps.${step_id}.outputs.${var}="${!var}"; echo "${var}=${!var}" >> $GITHUB_OUTPUT; done; }
+
+        TEST_SUMMARY_FILE='sed-gnu-result.json'
+        outputs TEST_SUMMARY_FILE
+
+    - name: Checkout code (sed)
+      uses: actions/checkout@v6
+      with:
+        path: 'sed'
+        persist-credentials: false
+
+    - name: Retrieve reference artifacts
+      uses: dawidd6/action-download-artifact@v12
+      continue-on-error: true
+      with:
+        workflow: GnuTests.yml
+        branch: "${{ env.DEFAULT_BRANCH }}"
+        workflow_conclusion: completed
+        path: "reference"
+        if_no_artifact_found: warn
+
+    - name: Download full json results
+      uses: actions/download-artifact@v7
+      with:
+        name: sed-gnu-full-result
+        path: results
+
+    - name: Extract/summarize testing info
+      id: summary
+      shell: bash
+      run: |
+        ## Extract/summarize testing info
+        outputs() { step_id="${{ github.action }}"; for var in "$@" ; do echo steps.${step_id}.outputs.${var}="${!var}"; echo "${var}=${!var}" >> $GITHUB_OUTPUT; done; }
+
+        # Check if results directory exists and has JSON files
+        json_count=0
+        if [[ -d "results" ]]; then
+          json_count=$(find results -name "*.json" | wc -l)
+        fi
+
+        if [[ "$json_count" -lt 1 ]]; then
+          echo "::error ::Failed to download results json files; failing early"
+          echo "::error ::Contents of results directory:"
+          ls -lR results || echo "::error ::Results directory does not exist"
+          exit 1
+        fi
+
+        # Extract summary from JSON results
+        RESULT_FILE="results/${{ env.TEST_FULL_SUMMARY_FILE }}"
+        if [[ -f "$RESULT_FILE" ]]; then
+          TOTAL=$(jq -r '.summary.total // 0' "$RESULT_FILE")
+          PASS=$(jq -r '.summary.passed // 0' "$RESULT_FILE")
+          FAIL=$(jq -r '.summary.failed // 0' "$RESULT_FILE")
+          SKIP=$(jq -r '.summary.skipped // 0' "$RESULT_FILE")
+          ERROR=0  # Our format doesn't distinguish errors from failures
+        else
+          echo "::error ::Result file $RESULT_FILE not found"
+          echo "::error ::Available files in results:"
+          find results -type f || true
+          TOTAL=0; PASS=0; FAIL=0; SKIP=0; ERROR=0
+        fi
+
+        output="GNU sed tests summary = TOTAL: $TOTAL / PASS: $PASS / FAIL: $FAIL / SKIP: $SKIP"
+        echo "${output}"
+
+        if [[ "$FAIL" -gt 0 ]]; then
+          echo "::warning ::${output}"
+        fi
+
+        jq -n \
+              --arg date "$(date --rfc-email)" \
+              --arg sha "$GITHUB_SHA" \
+              --arg total "$TOTAL" \
+              --arg pass "$PASS" \
+              --arg skip "$SKIP" \
+              --arg fail "$FAIL" \
+              --arg error "$ERROR" \
+              '{($date): { sha: $sha, total: $total, pass: $pass, skip: $skip, fail: $fail, error: $error }}' > '${{ steps.vars.outputs.TEST_SUMMARY_FILE }}'
+
+        HASH=$(sha1sum '${{ steps.vars.outputs.TEST_SUMMARY_FILE }}' | cut --delim=" " -f 1)
+        outputs HASH TOTAL PASS FAIL SKIP
+
+    - name: Upload SHA1/ID of 'test-summary'
+      uses: actions/upload-artifact@v6
+      with:
+        name: "${{ steps.summary.outputs.HASH }}"
+        path: "${{ steps.vars.outputs.TEST_SUMMARY_FILE }}"
+
+    - name: Upload test results summary
+      uses: actions/upload-artifact@v6
+      with:
+        name: test-summary
+        path: "${{ steps.vars.outputs.TEST_SUMMARY_FILE }}"
+
+    - name: Compare test failures VS reference
+      shell: bash
+      run: |
+        ## Compare test failures VS reference using JSON files
+        REF_SUMMARY_FILE='reference/sed-gnu-full-result/sed-gnu-full-result.json'
+        CURRENT_SUMMARY_FILE="results/${{ env.TEST_FULL_SUMMARY_FILE }}"
+        REPO_DEFAULT_BRANCH='${{ env.DEFAULT_BRANCH }}'
+
+        # Path to ignore file for intermittent issues
+        IGNORE_INTERMITTENT="sed/.github/workflows/ignore-intermittent.txt"
+
+        # Set up comment directory
+        COMMENT_DIR="reference/comment"
+        mkdir -p ${COMMENT_DIR}
+        echo ${{ github.event.number }} > ${COMMENT_DIR}/NR
+        COMMENT_LOG="${COMMENT_DIR}/result.txt"
+
+        COMPARISON_RESULT=0
+        if test -f "${CURRENT_SUMMARY_FILE}"; then
+          if test -f "${REF_SUMMARY_FILE}"; then
+            echo "Reference summary SHA1/ID: $(sha1sum -- "${REF_SUMMARY_FILE}")"
+            echo "Current summary SHA1/ID: $(sha1sum -- "${CURRENT_SUMMARY_FILE}")"
+
+            python3 sed/util/compare_test_results.py \
+              --ignore-file "${IGNORE_INTERMITTENT}" \
+              --output "${COMMENT_LOG}" \
+              "${CURRENT_SUMMARY_FILE}" "${REF_SUMMARY_FILE}"
+
+            COMPARISON_RESULT=$?
+          else
+            echo "::warning ::Skipping test comparison; no prior reference summary is available at '${REF_SUMMARY_FILE}'."
+          fi
+        else
+          echo "::error ::Failed to find summary of test results (missing '${CURRENT_SUMMARY_FILE}'); failing early"
+          exit 1
+        fi
+
+        if [ ${COMPARISON_RESULT} -eq 1 ]; then
+          echo "ONLY_INTERMITTENT=false" >> $GITHUB_ENV
+          echo "::error ::Found new non-intermittent test failures"
+          exit 1
+        else
+          echo "ONLY_INTERMITTENT=true" >> $GITHUB_ENV
+          echo "::notice ::No new test failures detected"
+        fi
+    
+    - name: Upload comparison log (for GnuComment workflow)
+      if: success() || failure()
+      uses: actions/upload-artifact@v6
+      with:
+        name: comment
+        path: reference/comment/
+
+    - name: Report test results
+      if: success() || failure()
+      shell: bash
+      run: |
+        ## Report final results
+        echo "::notice ::GNU sed testsuite results:"
+        echo "::notice ::  Total tests: ${{ steps.summary.outputs.TOTAL }}"
+        echo "::notice ::  Passed: ${{ steps.summary.outputs.PASS }}"
+        echo "::notice ::  Failed: ${{ steps.summary.outputs.FAIL }}"
+        echo "::notice ::  Skipped: ${{ steps.summary.outputs.SKIP }}"
+
+        if [[ "${{ steps.summary.outputs.FAIL }}" -gt 0 ]]; then
+          PASS_RATE=$(( ${{ steps.summary.outputs.PASS }} * 100 / (${{ steps.summary.outputs.PASS }} + ${{ steps.summary.outputs.FAIL }}) ))
+          echo "::notice ::  Pass rate: ${PASS_RATE}%"
+        fi

.github/workflows/code-quality.yml

@@ -69,9 +69,8 @@ jobs:
         run: |
           ## `cargo clippy` lint testing
           unset fault
-          CLIPPY_FLAGS="-W clippy::default_trait_access -W clippy::manual_string_new -W clippy::cognitive_complexity -W clippy::implicit_clone -W clippy::range-plus-one -W clippy::redundant-clone -W clippy::match_bool -W clippy::semicolon_if_nothing_returned"
           fault_type="${{ steps.vars.outputs.FAULT_TYPE }}"
           fault_prefix=$(echo "$fault_type" | tr '[:lower:]' '[:upper:]')
           # * convert any warnings to GHA UI annotations; ref: <https://help.github.com/en/actions/reference/workflow-commands-for-github-actions#setting-a-warning-message>
-          S=$(cargo clippy --all-targets --workspace -psed -- ${CLIPPY_FLAGS} -D warnings 2>&1) && printf "%s\n" "$S" || { printf "%s\n" "$S" ; printf "%s" "$S" | sed -E -n -e '/^error:/{' -e "N; s/^error:[[:space:]]+(.*)\\n[[:space:]]+-->[[:space:]]+(.*):([0-9]+):([0-9]+).*$/::${fault_type} file=\2,line=\3,col=\4::${fault_prefix}: \`cargo clippy\`: \1 (file:'\2', line:\3)/p;" -e '}' ; fault=true ; }
+          S=$(cargo clippy --all-targets --workspace -psed -- -D warnings 2>&1) && printf "%s\n" "$S" || { printf "%s\n" "$S" ; printf "%s" "$S" | sed -E -n -e '/^error:/{' -e "N; s/^error:[[:space:]]+(.*)\\n[[:space:]]+-->[[:space:]]+(.*):([0-9]+):([0-9]+).*$/::${fault_type} file=\2,line=\3,col=\4::${fault_prefix}: \`cargo clippy\`: \1 (file:'\2', line:\3)/p;" -e '}' ; fault=true ; }
           if [ -n "${{ steps.vars.outputs.FAIL_ON_FAULT }}" ] && [ -n "$fault" ]; then exit 1 ; fi

.github/workflows/ignore-intermittent.txt

@@ -0,0 +1,7 @@
+# List of intermittent test names to ignore in result comparisons
+# Format: one test name per line, lines starting with # are comments
+#
+# Add test names that are known to be flaky or environment-dependent
+# Example:
+# basic_substitution
+# line_address_test

Cargo.lock

@@ -56,7 +56,7 @@ version = "1.1.5"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "40c48f72fd53cd289104fc64099abca73db4166ad86ea0b4341abe65af83dadc"
 dependencies = [
- "windows-sys 0.61.2",
+ "windows-sys 0.60.2",
 ]
 
 [[package]]
@@ -67,7 +67,7 @@ checksum = "291e6a250ff86cd4a820112fb8898808a366d8f9f58ce16d1f538353ad55747d"
 dependencies = [
  "anstyle",
  "once_cell_polyfill",
- "windows-sys 0.61.2",
+ "windows-sys 0.60.2",
 ]
 
 [[package]]
@@ -127,6 +127,15 @@ version = "2.10.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "812e12b5285cc515a9c72a5c1d3b6d46a19dac5acfef5265968c166106e31dd3"
 
+[[package]]
+name = "block-buffer"
+version = "0.10.4"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "3078c7629b62d3f0439517fa394996acacc5cbc91c5a20d8c658e77abd503a71"
+dependencies = [
+ "generic-array",
+]
+
 [[package]]
 name = "bstr"
 version = "1.12.1"
@@ -167,9 +176,9 @@ checksum = "613afe47fcd5fac7ccf1db93babcb082c5994d996f20b8b159f2ad1658eb5724"
 
 [[package]]
 name = "chrono"
-version = "0.4.42"
+version = "0.4.43"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "145052bdd345b87320e369255277e3fb5152762ad123a901ef5c262dd38fe8d2"
+checksum = "fac4744fb15ae8337dc853fee7fb3f4e48c0fbaa23d0afe49c447b4fab126118"
 dependencies = [
  "iana-time-zone",
  "num-traits",
@@ -178,18 +187,18 @@ dependencies = [
 
 [[package]]
 name = "clap"
-version = "4.5.53"
+version = "4.5.55"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "c9e340e012a1bf4935f5282ed1436d1489548e8f72308207ea5df0e23d2d03f8"
+checksum = "3e34525d5bbbd55da2bb745d34b36121baac88d07619a9a09cfcf4a6c0832785"
 dependencies = [
  "clap_builder",
 ]
 
 [[package]]
 name = "clap_builder"
-version = "4.5.53"
+version = "4.5.55"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "d76b5d13eaa18c901fd2f7fca939fefe3a0727a953561fefdf3b2922b8569d00"
+checksum = "59a20016a20a3da95bef50ec7238dbd09baeef4311dcdd38ec15aba69812fb61"
 dependencies = [
  "anstream",
  "anstyle",
@@ -200,9 +209,9 @@ dependencies = [
 
 [[package]]
 name = "clap_complete"
-version = "4.5.61"
+version = "4.5.65"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "39615915e2ece2550c0149addac32fb5bd312c657f43845bb9088cb9c8a7c992"
+checksum = "430b4dc2b5e3861848de79627b2bedc9f3342c7da5173a14eaa5d0f8dc18ae5d"
 dependencies = [
  "clap",
 ]
@@ -225,9 +234,9 @@ dependencies = [
 
 [[package]]
 name = "codspeed"
-version = "4.1.0"
+version = "4.3.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "c3b847e05a34be5c38f3f2a5052178a3bd32e6b5702f3ea775efde95c483a539"
+checksum = "38c2eb3388ebe26b5a0ab6bf4969d9c4840143d7f6df07caa3cc851b0606cef6"
 dependencies = [
  "anyhow",
  "cc",
@@ -243,9 +252,9 @@ dependencies = [
 
 [[package]]
 name = "codspeed-divan-compat"
-version = "4.1.0"
+version = "4.3.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "f0f0e9fe5eaa39995ec35e46407f7154346cc25bd1300c64c21636f3d00cb2cc"
+checksum = "b2de65b7489a59709724d489070c6d05b7744039e4bf751d0a2006b90bb5593d"
 dependencies = [
  "clap",
  "codspeed",
@@ -256,9 +265,9 @@ dependencies = [
 
 [[package]]
 name = "codspeed-divan-compat-macros"
-version = "4.1.0"
+version = "4.3.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "88c8babf2a40fd2206a2e030cf020d0d58144cd56e1dc408bfba02cdefb08b4f"
+checksum = "56ca01ce4fd22b8dcc6c770dcd6b74343642e842482b94e8920d14e10c57638d"
 dependencies = [
  "divan-macros",
  "itertools",
@@ -270,9 +279,9 @@ dependencies = [
 
 [[package]]
 name = "codspeed-divan-compat-walltime"
-version = "4.1.0"
+version = "4.3.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "7f26092328e12a36704ffc552f379c6405dd94d3149970b79b22d371717c2aae"
+checksum = "720ab9d0714718afe5f5832be6e5f5eb5ce97836e24ca7bf7042eea4308b9fb8"
 dependencies = [
  "cfg-if",
  "clap",
@@ -311,6 +320,15 @@ version = "0.8.7"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "773648b94d0e5d620f64f280777445740e61fe701025087ec8b57f45c791888b"
 
+[[package]]
+name = "cpufeatures"
+version = "0.2.17"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "59ed5838eebb26a2bb2e58f6d5b5316989ae9d08bab10e0e6d103e656d1b0280"
+dependencies = [
+ "libc",
+]
+
 [[package]]
 name = "crossbeam-deque"
 version = "0.8.6"
@@ -336,6 +354,16 @@ version = "0.8.21"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "d0a5c400df2834b80a4c3327b3aad3a4c4cd4de0629063962b03235697506a28"
 
+[[package]]
+name = "crypto-common"
+version = "0.1.7"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "78c8292055d1c1df0cce5d180393dc8cce0abec0a7102adb6c7b1eef6016d60a"
+dependencies = [
+ "generic-array",
+ "typenum",
+]
+
 [[package]]
 name = "ctor"
 version = "0.6.3"
@@ -373,6 +401,16 @@ version = "0.4.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "6184e33543162437515c2e2b48714794e37845ec9851711914eec9d308f6ebe8"
 
+[[package]]
+name = "digest"
+version = "0.10.7"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "9ed9a281f7bc9b7576e61468ba615a66a5c8cfdff42420a70aa82701a3b1e292"
+dependencies = [
+ "block-buffer",
+ "crypto-common",
+]
+
 [[package]]
 name = "displaydoc"
 version = "0.2.5"
@@ -447,7 +485,7 @@ source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "39cab71617ae0d63f51a36d69f866391735b51691dbda63cf6f96d042b63efeb"
 dependencies = [
  "libc",
- "windows-sys 0.61.2",
+ "windows-sys 0.59.0",
 ]
 
 [[package]]
@@ -527,6 +565,16 @@ dependencies = [
  "thiserror",
 ]
 
+[[package]]
+name = "generic-array"
+version = "0.14.7"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "85649ca51fd72272d7821adaf274ad91c288277713d9c18820d8499a7ff69e9a"
+dependencies = [
+ "typenum",
+ "version_check",
+]
+
 [[package]]
 name = "getrandom"
 version = "0.2.16"
@@ -586,6 +634,12 @@ version = "0.16.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "841d1cc9bed7f9236f321df977030373f4a4163ae1a7dbfe1a51a2c1a51d9100"
 
+[[package]]
+name = "hex"
+version = "0.4.3"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "7f24254aa9a54b5c858eaee2f5bccdb46aaf0e486a595ed5fd8f86ba55232a70"
+
 [[package]]
 name = "iana-time-zone"
 version = "0.1.64"
@@ -598,7 +652,7 @@ dependencies = [
  "js-sys",
  "log",
  "wasm-bindgen",
- "windows-core 0.62.2",
+ "windows-core 0.61.2",
 ]
 
 [[package]]
@@ -688,7 +742,7 @@ dependencies = [
  "portable-atomic",
  "portable-atomic-util",
  "serde_core",
- "windows-sys 0.61.2",
+ "windows-sys 0.59.0",
 ]
 
 [[package]]
@@ -735,9 +789,9 @@ checksum = "bbd2bcb4c963f2ddae06a2efc7e9f3591312473c50c6685e1f298068316e66fe"
 
 [[package]]
 name = "libc"
-version = "0.2.178"
+version = "0.2.180"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "37c93d8daa9d8a012fd8ab92f088405fb202ea0b6ab73ee2482ae66af4f42091"
+checksum = "bcc35a38544a891a5f7c865aca548a982ccb3b8650a5b06d0fd33a10283c56fc"
 
 [[package]]
 name = "linux-raw-sys"
@@ -1086,15 +1140,15 @@ checksum = "357703d41365b4b27c590e3ed91eabb1b663f07c4c084095e60cbed4362dff0d"
 
 [[package]]
 name = "rustix"
-version = "1.1.2"
+version = "1.1.3"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "cd15f8a2c5551a84d56efdc1cd049089e409ac19a3072d5037a17fd70719ff3e"
+checksum = "146c9e247ccc180c1f61615433868c99f3de3ae256a30a43b49f67c2d9171f34"
 dependencies = [
  "bitflags",
  "errno",
  "libc",
  "linux-raw-sys",
- "windows-sys 0.61.2",
+ "windows-sys 0.59.0",
 ]
 
 [[package]]
@@ -1130,10 +1184,10 @@ dependencies = [
  "codspeed-divan-compat",
  "ctor",
  "fancy-regex",
+ "hex",
  "libc",
  "memchr",
  "memmap2",
- "once_cell",
  "phf",
  "phf_codegen",
  "predicates",
@@ -1141,6 +1195,7 @@ dependencies = [
  "rand",
  "regex",
  "rlimit",
+ "sha2",
  "sysinfo",
  "tempfile",
  "terminal_size",
@@ -1199,6 +1254,17 @@ dependencies = [
  "serde_core",
 ]
 
+[[package]]
+name = "sha2"
+version = "0.10.9"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "a7507d819769d01a365ab707794a4084392c824f54a7a6a7862f8c3d0892b283"
+dependencies = [
+ "cfg-if",
+ "cpufeatures",
+ "digest",
+]
+
 [[package]]
 name = "shlex"
 version = "1.3.0"
@@ -1262,9 +1328,9 @@ dependencies = [
 
 [[package]]
 name = "sysinfo"
-version = "0.37.2"
+version = "0.38.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "16607d5caffd1c07ce073528f9ed972d88db15dd44023fa57142963be3feb11f"
+checksum = "fe840c5b1afe259a5657392a4dbb74473a14c8db999c3ec2f4ae812e028a94da"
 dependencies = [
  "libc",
  "memchr",
@@ -1276,15 +1342,15 @@ dependencies = [
 
 [[package]]
 name = "tempfile"
-version = "3.23.0"
+version = "3.24.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "2d31c77bdf42a745371d260a26ca7163f1e0924b64afa0b688e61b5a9fa02f16"
+checksum = "655da9c7eb6305c55742045d5a8d2037996d61d8de95806335c7c86ce0f82e9c"
 dependencies = [
  "fastrand",
  "getrandom 0.3.4",
  "once_cell",
  "rustix",
- "windows-sys 0.61.2",
+ "windows-sys 0.59.0",
 ]
 
 [[package]]
@@ -1418,6 +1484,12 @@ dependencies = [
  "rustc-hash",
 ]
 
+[[package]]
+name = "typenum"
+version = "1.19.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "562d481066bde0658276a35467c4af00bdc6ee726305698a55b86e61d7ad82bb"
+
 [[package]]
 name = "unic-langid"
 version = "0.9.6"
@@ -1513,6 +1585,12 @@ dependencies = [
  "xattr",
 ]
 
+[[package]]
+name = "version_check"
+version = "0.9.5"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "0b928f33d975fc6ad9f86c8f283853ad26bdd5b10b7f1542aa2fa15e2289105a"
+
 [[package]]
 name = "walkdir"
 version = "2.5.0"
@@ -1614,7 +1692,7 @@ version = "0.1.11"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "c2a7b1c03c876122aa43f3020e6c3c3ee5c05081c9a00739faf7503aeba10d22"
 dependencies = [
- "windows-sys 0.61.2",
+ "windows-sys 0.59.0",
 ]
 
 [[package]]
@@ -1625,24 +1703,23 @@ checksum = "712e227841d057c1ee1cd2fb22fa7e5a5461ae8e48fa2ca79ec42cfc1931183f"
 
 [[package]]
 name = "windows"
-version = "0.61.3"
+version = "0.62.2"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "9babd3a767a4c1aef6900409f85f5d53ce2544ccdfaa86dad48c91782c6d6893"
+checksum = "527fadee13e0c05939a6a05d5bd6eec6cd2e3dbd648b9f8e447c6518133d8580"
 dependencies = [
  "windows-collections",
- "windows-core 0.61.2",
+ "windows-core 0.62.2",
  "windows-future",
- "windows-link 0.1.3",
  "windows-numerics",
 ]
 
 [[package]]
 name = "windows-collections"
-version = "0.2.0"
+version = "0.3.2"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "3beeceb5e5cfd9eb1d76b381630e82c4241ccd0d27f1a39ed41b2760b255c5e8"
+checksum = "23b2d95af1a8a14a3c7367e1ed4fc9c20e0a26e79551b1454d72583c97cc6610"
 dependencies = [
- "windows-core 0.61.2",
+ "windows-core 0.62.2",
 ]
 
 [[package]]
@@ -1673,12 +1750,12 @@ dependencies = [
 
 [[package]]
 name = "windows-future"
-version = "0.2.1"
+version = "0.3.2"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "fc6a41e98427b19fe4b73c550f060b59fa592d7d686537eebf9385621bfbad8e"
+checksum = "e1d6f90251fe18a279739e78025bd6ddc52a7e22f921070ccdc67dde84c605cb"
 dependencies = [
- "windows-core 0.61.2",
+ "windows-core 0.62.2",
- "windows-link 0.1.3",
+ "windows-link 0.2.1",
  "windows-threading",
 ]
 
@@ -1718,12 +1795,12 @@ checksum = "f0805222e57f7521d6a62e36fa9163bc891acd422f971defe97d64e70d0a4fe5"
 
 [[package]]
 name = "windows-numerics"
-version = "0.2.0"
+version = "0.3.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "9150af68066c4c5c07ddc0ce30421554771e528bde427614c61038bc2c92c2b1"
+checksum = "6e2e40844ac143cdb44aead537bbf727de9b044e107a0f1220392177d15b0f26"
 dependencies = [
- "windows-core 0.61.2",
+ "windows-core 0.62.2",
- "windows-link 0.1.3",
+ "windows-link 0.2.1",
 ]
 
 [[package]]
@@ -1780,15 +1857,6 @@ dependencies = [
  "windows-targets 0.53.5",
 ]
 
-[[package]]
-name = "windows-sys"
-version = "0.61.2"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "ae137229bcbd6cdf0f7b80a31df61766145077ddf49416a728b02cb3921ff3fc"
-dependencies = [
- "windows-link 0.2.1",
-]
-
 [[package]]
 name = "windows-targets"
 version = "0.52.6"
@@ -1824,11 +1892,11 @@ dependencies = [
 
 [[package]]
 name = "windows-threading"
-version = "0.1.0"
+version = "0.2.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "b66463ad2e0ea3bbf808b7f1d371311c80e115c0b71d60efc142cafbcfb057a6"
+checksum = "3949bd5b99cafdf1c7ca86b43ca564028dfe27d66958f2470940f73d86d75b37"
 dependencies = [
- "windows-link 0.1.3",
+ "windows-link 0.2.1",
 ]
 
 [[package]]

Cargo.toml

@@ -38,16 +38,17 @@ clap_complete = "4.5"
 clap_mangen = "0.2"
 divan = { package = "codspeed-divan-compat", version = "4.0.5" }
 fancy-regex = "0.17.0"
+hex = "0.4"
 libc = "0.2.153"
 memchr = "2.7.4"
 memmap2 = "0.9"
-once_cell = "1.21.3"
 phf = "0.13.0"
 phf_codegen = "0.13.0"
 predicates = "3.1.3"
 rand = { version = "0.9", features = ["small_rng"] }
 regex = "1.10.4"
-sysinfo = "0.37"
+sha2 = "0.10"
+sysinfo = "0.38"
 tempfile = "3.10.1"
 textwrap = { version = "0.16.1", features = ["terminal_size"] }
 terminal_size = "0.4.2"
@@ -64,7 +65,6 @@ ctor = "0.6.0"
 fancy-regex =  { workspace = true }
 memchr = { workspace = true }
 memmap2.workspace = true
-once_cell = { workspace = true }
 phf = { workspace = true }
 predicates = { workspace = true }
 regex = { workspace = true }
@@ -77,10 +77,12 @@ uucore = { workspace = true }
 [dev-dependencies]
 chrono = { workspace = true }
 divan = { workspace = true }
+hex = { workspace = true }
 libc = { workspace = true }
 pretty_assertions = "1"
 rand = { workspace = true }
 regex = { workspace = true }
+sha2 = { workspace = true }
 tempfile = { workspace = true }
 uucore = { workspace = true, features = ["entries", "process", "signals", "benchmark"] }
 uutests = "0.5.0"
@@ -116,7 +118,7 @@ lto = true
 [profile.release-fast]
 inherits = "release"
 panic = "abort"
-codegen-units = 1
+codegen-units = 1 # should be moved to release without regression
 
 # A release-like profile that is as small as possible.
 [profile.release-small]
@@ -124,7 +126,12 @@ inherits = "release-fast"
 opt-level = "z"
 strip = true
 
-# The profile that 'cargo dist' will build with
+[lints.clippy]
-[profile.dist]
+default_trait_access = "warn"
-inherits = "release"
+manual_string_new = "warn"
-lto = "thin"
+cognitive_complexity = "warn"
+implicit_clone = "warn"
+range-plus-one = "warn"
+redundant-clone = "warn"
+match_bool = "warn"
+semicolon_if_nothing_returned = "warn"

README.md

@@ -41,6 +41,31 @@ cargo run --release
 
 The binary is named `sed` in `target/release/sed`.
 
+## Testing
+
+### GNU sed Compatibility Testing
+
+Test compatibility against GNU sed using the comprehensive testsuite (47+ tests, ~10% pass rate):
+
+```bash
+# Clone GNU sed testsuite (one time setup)
+git clone https://github.com/mirror/sed.git ../gnu.sed
+
+# Run compatibility tests
+./util/run-gnu-testsuite.sh
+
+# Generate JSON results for CI
+./util/run-gnu-testsuite.sh --json-output results.json
+```
+
+The testsuite extracts test cases from the GNU sed repository and tests them against expected outputs.
+
+### Unit Tests
+
+```bash
+cargo test
+```
+
 ## Extensions and incompatibilities
 ### Supported GNU extensions
 * Command-line arguments can be specified in long (`--`) form.
@@ -58,6 +83,10 @@ The binary is named `sed` in `target/release/sed`.
 * The `q` command can be optionally followed by an exit code.
 * The `l` command can be optionally followed by the output width.
 * The `--follow-symlinks` flag for in-place editing.
+* Address 0 can be used to specify an address range that is already
+  active on line 1 and can finish with the specified regular expression.
+* Address steps can be specified in the form of start~step and start,~step
+  ranges.
 
 ### Supported BSD and GNU extensions
 * The second address in a range can be specified as a relative address with +N.
@@ -87,6 +116,13 @@ The binary is named `sed` in `target/release/sed`.
 * Labels are parsed for alphanumeric characters. The BSD version parses them
   until the end of the line, preventing ; to be used as a separator.
 
+## GNU test suite compatibility
+
+Below is the evolution of how many GNU tests uutils passes.
+
+![Evolution over time](https://github.com/uutils/sed-tracking/blob/main/gnu-results.svg?raw=true)
+
+
 ## License
 
 sed is licensed under the MIT License - see the `LICENSE` file for details

benches/sed_bench.rs

@@ -17,7 +17,7 @@ fn no_op_short(bencher: Bencher) {
     let temp_dir = tempfile::tempdir().unwrap();
     let mut data = Vec::new();
     for i in 0..10_000_000 {
-        data.extend_from_slice(format!("{}\n", i).as_bytes());
+        data.extend_from_slice(format!("{i}\n").as_bytes());
     }
     let file_path = create_test_file(&data, temp_dir.path());
     let file_path_str = file_path.to_str().unwrap();
@@ -192,7 +192,7 @@ fn remove_cr(bencher: Bencher) {
     let temp_dir = tempfile::tempdir().unwrap();
     let mut data = Vec::new();
     for i in 0..500_000 {
-        let line = format!("line {} with windows endings\r\n", i);
+        let line = format!("line {i} with windows endings\r\n");
         data.extend_from_slice(line.as_bytes());
     }
     let file_path = create_test_file(&data, temp_dir.path());
@@ -240,7 +240,7 @@ fn number_fix(bencher: Bencher) {
         let cents = i % 100;
         let thousands = euros / 1000;
         let remainder = euros % 1000;
-        let line = format!("{}.{:03},{:02}\n", thousands, remainder, cents);
+        let line = format!("{thousands}.{remainder:03},{cents:02}\n");
         data.extend_from_slice(line.as_bytes());
     }
     let file_path = create_test_file(&data, temp_dir.path());

fuzz/Cargo.lock

@@ -579,9 +579,9 @@ dependencies = [
 
 [[package]]
 name = "libc"
-version = "0.2.178"
+version = "0.2.180"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "37c93d8daa9d8a012fd8ab92f088405fb202ea0b6ab73ee2482ae66af4f42091"
+checksum = "bcc35a38544a891a5f7c865aca548a982ccb3b8650a5b06d0fd33a10283c56fc"
 
 [[package]]
 name = "libfuzzer-sys"
@@ -601,9 +601,9 @@ checksum = "f9fbbcab51052fe104eb5e5d351cf728d30a5be1fe14d9be8a3b097481fb97de"
 
 [[package]]
 name = "linux-raw-sys"
-version = "0.9.4"
+version = "0.11.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "cd945864f07fe9f5371a27ad7b52a172b4b499999f1d97574c9fa68373937e12"
+checksum = "df1d3c3b53da64cf5760482273a98e575c651a67eec7f77df96b5b642de8f039"
 
 [[package]]
 name = "log"
@@ -923,15 +923,15 @@ checksum = "357703d41365b4b27c590e3ed91eabb1b663f07c4c084095e60cbed4362dff0d"
 
 [[package]]
 name = "rustix"
-version = "1.0.7"
+version = "1.1.3"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "c71e83d6afe7ff64890ec6b71d6a69bb8a610ab78ce364b3352876bb4c801266"
+checksum = "146c9e247ccc180c1f61615433868c99f3de3ae256a30a43b49f67c2d9171f34"
 dependencies = [
  "bitflags",
  "errno",
  "libc",
  "linux-raw-sys",
- "windows-sys 0.59.0",
+ "windows-sys 0.61.0",
 ]
 
 [[package]]
@@ -951,7 +951,7 @@ dependencies = [
 
 [[package]]
 name = "sed"
-version = "0.0.1"
+version = "0.1.1"
 dependencies = [
  "assert_fs",
  "clap",
@@ -961,7 +961,6 @@ dependencies = [
  "fancy-regex",
  "memchr",
  "memmap2",
- "once_cell",
  "phf",
  "phf_codegen",
  "predicates",
@@ -1068,15 +1067,15 @@ dependencies = [
 
 [[package]]
 name = "tempfile"
-version = "3.23.0"
+version = "3.24.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "2d31c77bdf42a745371d260a26ca7163f1e0924b64afa0b688e61b5a9fa02f16"
+checksum = "655da9c7eb6305c55742045d5a8d2037996d61d8de95806335c7c86ce0f82e9c"
 dependencies = [
  "fastrand",
  "getrandom",
  "once_cell",
  "rustix",
- "windows-sys 0.59.0",
+ "windows-sys 0.61.0",
 ]
 
 [[package]]

src/sed/command.rs

@@ -64,6 +64,8 @@ pub struct ProcessingContext {
     pub parsed_block_nesting: usize,
     /// Command associated with each label
     pub label_to_command_map: HashMap<String, Rc<RefCell<Command>>>,
+    /// Commands with a (latchable and resetable) address range
+    pub range_commands: Vec<Rc<RefCell<Command>>>,
     /// True if a substitution was made as specified in the t command
     pub substitution_made: bool,
     /// Elements to append at the end of each command processing cycle
@@ -84,26 +86,15 @@ pub struct StringSpace {
     pub has_newline: bool, // True if \n-terminated
 }
 
-#[derive(Debug, Clone, Copy, PartialEq, Eq)]
+#[derive(Debug)]
 /// Types of address specifications that precede commands
-pub enum AddressType {
+pub enum Address {
-    Re,      // Line that matches regex
+    Re(Option<Regex>), // Line that matches (optional) regex
-    Line,    // Specific line
+    Line(usize),       // Specific line
-    RelLine, // Relative line
+    RelLine(usize),    // Relative line
-    Last,    // Last line
+    Last,              // Last line
-}
+    StepMatch(usize),  // Lines matching specified step from first
-
+    StepEnd(usize),    // Range ending at specified step from first
-#[derive(Debug)]
-/// Format of an address
-pub struct Address {
-    pub atype: AddressType,  // Address type
-    pub value: AddressValue, // Line number or regex
-}
-
-#[derive(Debug)]
-pub enum AddressValue {
-    LineNumber(usize),
-    Regex(Option<Regex>),
 }
 
 #[derive(Debug)]
@@ -278,7 +269,7 @@ pub struct Command {
     pub addr1: Option<Address>,             // Start address
     pub addr2: Option<Address>,             // End address
     pub non_select: bool,                   // True if '!'
-    pub start_line: Option<usize>,          // Start line number (or None)
+    pub start_line: Option<usize>,          // Start line number (or None if unlatched)
     pub data: CommandData,                  // Command-specific data
     pub next: Option<Rc<RefCell<Command>>>, // Pointer to next command
     pub location: ScriptLocation,           // Command's definition location

src/sed/fast_io.rs

@@ -61,7 +61,6 @@ pub struct MmapLineCursor<'a> {
 pub struct NextMmapLine<'a> {
     pub content: &'a [u8],
     pub full_span: &'a [u8],
-    pub is_last_line: bool,
 }
 
 #[cfg(unix)]
@@ -76,7 +75,7 @@ impl<'a> MmapLineCursor<'a> {
     }
 
     /// Return the next line, if available, or None.
-    fn get_line(&mut self) -> io::Result<Option<NextMmapLine<'_>>> {
+    fn get_line(&mut self) -> io::Result<Option<NextMmapLine<'a>>> {
         if self.pos >= self.data.len() {
             return Ok(None);
         }
@@ -101,12 +100,12 @@ impl<'a> MmapLineCursor<'a> {
             full_span
         };
 
-        let is_last_line = self.pos >= self.data.len();
+        Ok(Some(NextMmapLine { content, full_span }))
-        Ok(Some(NextMmapLine {
+    }
-            content,
+
-            full_span,
+    /// Return true if the previously returned line was the last one.
-            is_last_line,
+    fn last_line(&mut self) -> io::Result<bool> {
-        }))
+        Ok(self.pos >= self.data.len())
     }
 }
 
@@ -126,9 +125,8 @@ impl ReadLineCursor {
         }
     }
 
-    /// If a line is available, return it, its \n termination,
+    /// If a line is available, return it and its \n termination.
-    /// and next line availability, otherwise return None.
+    fn get_line(&mut self) -> io::Result<Option<(String, bool)>> {
-    fn get_line(&mut self) -> io::Result<Option<(String, bool, bool)>> {
         self.buffer.clear();
         // read_line *includes* the '\n' if present
         let bytes_read = self.reader.read_line(&mut self.buffer)?;
@@ -142,8 +140,14 @@ impl ReadLineCursor {
             self.buffer.pop();
         }
         let line = std::mem::take(&mut self.buffer);
-        let is_last_line = self.reader.fill_buf()?.is_empty();
+        Ok(Some((line, has_newline)))
-        Ok(Some((line, has_newline, is_last_line)))
+    }
+
+    /// Return true if the previously returned line was the last one.
+    fn last_line(&mut self) -> io::Result<bool> {
+        // FIXME(rust-lang#86423): Replace with BufRead::has_data_left()
+        // when/if method becomes stable.
+        Ok(self.reader.fill_buf()?.is_empty())
     }
 }
 
@@ -466,21 +470,15 @@ impl<'a> LineReader<'a> {
         line_reader_read_input(file)
     }
 
-    /// Return the next line, if available and also the availability
+    /// Return the next line, if available.
-    /// of another one, or None at end of file.
+    pub fn get_line(&mut self) -> io::Result<Option<IOChunk<'a>>> {
-    pub fn get_line(&mut self) -> io::Result<Option<(IOChunk<'_>, bool)>> {
         match self {
             #[cfg(unix)]
             LineReader::MmapInput { cursor, .. } => {
                 // Obtain fields to prevent borrowing issues.
                 let fast_copy = cursor.fast_copy.clone();
                 let base = cursor.data.as_ptr();
-                if let Some(NextMmapLine {
+                if let Some(NextMmapLine { content, full_span }) = cursor.get_line()? {
-                    content,
-                    full_span,
-                    is_last_line,
-                }) = cursor.get_line()?
-                {
                     let chunk = IOChunk::from_content(IOChunkContent::MmapInput {
                         fast_copy,
                         base,
@@ -488,17 +486,17 @@ impl<'a> LineReader<'a> {
                         full_span,
                     });
 
-                    Ok(Some((chunk, is_last_line)))
+                    Ok(Some(chunk))
                 } else {
                     Ok(None)
                 }
             }
 
             LineReader::ReadInput(cursor) => {
-                if let Some((line, _has_newline, is_last_line)) = cursor.get_line()? {
+                if let Some((line, _has_newline)) = cursor.get_line()? {
                     let chunk =
                         IOChunk::from_content(IOChunkContent::new_owned(line, _has_newline));
-                    Ok(Some((chunk, is_last_line)))
+                    Ok(Some(chunk))
                 } else {
                     Ok(None)
                 }
@@ -508,6 +506,19 @@ impl<'a> LineReader<'a> {
             LineReader::_Phantom(_) => unreachable!("_Phantom should never be constructed"),
         }
     }
+
+    /// Return true if the previously returned line was the last one.
+    pub fn last_line(&mut self) -> io::Result<bool> {
+        match self {
+            #[cfg(unix)]
+            LineReader::MmapInput { cursor, .. } => cursor.last_line(),
+
+            LineReader::ReadInput(cursor) => cursor.last_line(),
+
+            #[cfg(not(unix))]
+            LineReader::_Phantom(_) => unreachable!("_Phantom should never be constructed"),
+        }
+    }
 }
 
 // Define a trait combining two: workaround for Rust's corresponding inability.
@@ -1048,7 +1059,7 @@ mod tests {
         let mut out = OutputBuffer::new(Box::new(Box::new(out_file)));
 
         // Drain reader → writer
-        while let Some((chunk, _last_line)) = reader.get_line()? {
+        while let Some(chunk) = reader.get_line()? {
             out.write_chunk(&chunk)?;
         }
         out.flush()?;
@@ -1084,7 +1095,7 @@ mod tests {
         let mut out = OutputBuffer::new(Box::new(out_file));
 
         // Read the first mmap line ("zero\n") and write it
-        if let Some((chunk, _last_line)) = reader.get_line()? {
+        if let Some(chunk) = reader.get_line()? {
             out.write_chunk(&chunk)?;
         }
 
@@ -1092,7 +1103,7 @@ mod tests {
         out.write_str("middle\n")?;
 
         // Read the second mmap line ("one\n") and write it
-        if let Some((chunk, _last_line)) = reader.get_line()? {
+        if let Some(chunk) = reader.get_line()? {
             out.write_chunk(&chunk)?;
         }
 
@@ -1137,7 +1148,7 @@ mod tests {
         // Wrap it in your OutputBuffer and run the loop:
         let mut out = OutputBuffer::new(Box::new(out_file));
         let mut nline = 0;
-        while let Some((chunk, _last_line)) = reader.get_line()? {
+        while let Some(chunk) = reader.get_line()? {
             out.write_chunk(&chunk)?;
             nline += 1;
         }
@@ -1176,7 +1187,7 @@ mod tests {
         // Wrap it in your OutputBuffer and run the loop:
         let mut out = OutputBuffer::new(Box::new(out_file));
         let mut nline = 0;
-        while let Some((chunk, _last_line)) = reader.get_line()? {
+        while let Some(chunk) = reader.get_line()? {
             out.write_chunk(&chunk)?;
             nline += 1;
         }
@@ -1211,7 +1222,7 @@ mod tests {
         // Wrap it in your OutputBuffer and run the loop:
         let mut out = OutputBuffer::new(Box::new(out_file));
         let mut nline = 0;
-        while let Some((chunk, _last_line)) = reader.get_line()? {
+        while let Some(chunk) = reader.get_line()? {
             out.write_chunk(&chunk)?;
             nline += 1;
         }
@@ -1246,7 +1257,7 @@ mod tests {
         // Wrap it in your OutputBuffer and run the loop:
         let mut out = OutputBuffer::new(Box::new(out_file));
         let mut nline = 0;
-        while let Some((chunk, _last_line)) = reader.get_line()? {
+        while let Some(chunk) = reader.get_line()? {
             out.write_chunk(&chunk)?;
             nline += 1;
         }
@@ -1288,7 +1299,7 @@ mod tests {
         // Wrap it in OutputBuffer and run the loop:
         let mut out = OutputBuffer::new(Box::new(out_file));
         let mut nline_written = 0;
-        while let Some((chunk, _last_line)) = reader.get_line()? {
+        while let Some(chunk) = reader.get_line()? {
             out.write_chunk(&chunk)?;
             nline_written += 1;
         }
@@ -1319,52 +1330,46 @@ mod tests {
         let mut reader = LineReader::open_stream(&path)?;
 
         // Verify the reader's operation
-        if let Some((
+        if let Some(IOChunk {
-            IOChunk {
+            content:
-                content:
+                IOChunkContent::Owned {
-                    IOChunkContent::Owned {
+                    content,
-                        content,
+                    has_newline,
-                        has_newline,
+                    ..
-                        ..
+                },
-                    },
+            utf8_verified,
-                utf8_verified,
+            ..
-                ..
+        }) = reader.get_line()?
-            },
-            last_line,
-        )) = reader.get_line()?
         {
             assert_eq!(content, "first line");
             assert_eq!(content.len(), 10);
             assert!(has_newline);
             assert!(!utf8_verified.get());
-            assert!(!last_line);
+            assert!(!reader.last_line().unwrap());
         } else {
             panic!("Expected IOChunkContent::Owned");
         }
 
-        if let Some((
+        if let Some(IOChunk {
-            IOChunk {
+            content:
-                content:
+                IOChunkContent::Owned {
-                    IOChunkContent::Owned {
+                    content,
-                        content,
+                    has_newline,
-                        has_newline,
+                    ..
-                        ..
+                },
-                    },
+            ..
-                ..
+        }) = reader.get_line()?
-            },
-            last_line,
-        )) = reader.get_line()?
         {
             assert_eq!(content, "second line");
             assert!(has_newline);
-            assert!(!last_line);
+            assert!(!reader.last_line().unwrap());
         } else {
             panic!("Expected IOChunkContent::Owned");
         }
 
-        if let Some((content, last_line)) = reader.get_line()? {
+        if let Some(content) = reader.get_line()? {
             assert_eq!(content.as_str().unwrap(), "last line");
-            assert!(last_line);
+            assert!(reader.last_line().unwrap());
         } else {
             panic!("Expected IOChunk");
         }
@@ -1386,52 +1391,46 @@ mod tests {
         let mut reader = LineReader::open(&path)?;
 
         // Verify the reader's operation
-        if let Some((
+        if let Some(IOChunk {
-            IOChunk {
+            content:
-                content:
+                IOChunkContent::MmapInput {
-                    IOChunkContent::MmapInput {
+                    content, full_span, ..
-                        content, full_span, ..
+                },
-                    },
+            utf8_verified,
-                utf8_verified,
+            ..
-                ..
+        }) = reader.get_line()?
-            },
-            last_line,
-        )) = reader.get_line()?
         {
             assert_eq!(content, b"first line");
             assert_eq!(content.len(), 10);
             assert_eq!(full_span, b"first line\n");
             assert!(!utf8_verified.get());
-            assert!(!last_line);
+            assert!(!reader.last_line().unwrap());
         } else {
             panic!("Expected IOChunkContent::MapInput");
         }
 
-        if let Some((
+        if let Some(IOChunk {
-            IOChunk {
+            content:
-                content:
+                IOChunkContent::MmapInput {
-                    IOChunkContent::MmapInput {
+                    content, full_span, ..
-                        content, full_span, ..
+                },
-                    },
+            utf8_verified,
-                utf8_verified,
+            ..
-                ..
+        }) = reader.get_line()?
-            },
-            last_line,
-        )) = reader.get_line()?
         {
             assert_eq!(content, b"second line");
             assert_eq!(full_span, b"second line\n");
             assert!(!utf8_verified.get());
-            assert!(!last_line);
+            assert!(!reader.last_line().unwrap());
         } else {
             panic!("Expected IOChunkContent::MapInput");
         }
 
-        if let Some((content, last_line)) = reader.get_line()? {
+        if let Some(content) = reader.get_line()? {
             assert_eq!(content.as_bytes(), b"last line");
             assert_eq!(content.as_str().unwrap(), "last line");
             assert!(content.utf8_verified.get());
-            assert!(last_line);
+            assert!(reader.last_line().unwrap());
             // Cached version
             assert_eq!(content.as_str().unwrap(), "last line");
         } else {

src/sed/fast_regex.rs

@@ -15,12 +15,12 @@ use fancy_regex::{
     CaptureMatches as FancyCaptureMatches, Captures as FancyCaptures, Regex as FancyRegex,
 };
 use memchr::memmem;
-use once_cell::sync::Lazy;
 use regex::Regex as RustRegex;
 use regex::bytes::{
     CaptureMatches as ByteCaptureMatches, Captures as ByteCaptures, Regex as ByteRegex,
 };
 use std::error::Error;
+use std::sync::LazyLock;
 use uucore::error::{UResult, USimpleError};
 
 use crate::sed::fast_io::IOChunk;
@@ -32,7 +32,8 @@ use crate::sed::fast_io::IOChunk;
 // For example, r"\\1" and r"[\1]" will match, whereas only a number
 // after an odd number of backslashes and outside a character class
 // should match.
-static NEEDS_FANCY_RE: Lazy<RustRegex> = Lazy::new(|| regex::Regex::new(r"\\[1-9]").unwrap());
+static NEEDS_FANCY_RE: LazyLock<RustRegex> =
+    LazyLock::new(|| regex::Regex::new(r"\\[1-9]").unwrap());
 
 /// All characters signifying that the match must be handled by an RE
 /// rather than by plain string pattern matching.
@@ -41,7 +42,7 @@ static NEEDS_FANCY_RE: Lazy<RustRegex> = Lazy::new(|| regex::Regex::new(r"\\[1-9
 // matching, because Regex always constructs an automaton and needs
 // to handle state transitions, whereas plain string matching can
 // use tailored CPU string or vectored instructions.
-static NEEDS_RE: Lazy<RustRegex> = Lazy::new(|| {
+static NEEDS_RE: LazyLock<RustRegex> = LazyLock::new(|| {
     regex::Regex::new(
         r"(?x) # Turn on verbose mode
           ( ^                   # Non-escaped: i.e. at BOL

src/sed/mod.rs

@@ -216,6 +216,7 @@ fn build_context(matches: &ArgMatches) -> ProcessingContext {
         hold: StringSpace::default(),
         parsed_block_nesting: 0,
         label_to_command_map: HashMap::new(),
+        range_commands: Vec::new(),
         substitution_made: false,
         append_elements: Vec::new(),
     }

src/sed/processor.rs

@@ -9,8 +9,7 @@
 // file that was distributed with this source code.
 
 use crate::sed::command::{
-    Address, AddressType, AddressValue, AppendElement, Command, CommandData, InputAction,
+    Address, AppendElement, Command, CommandData, InputAction, ProcessingContext, Transliteration,
-    ProcessingContext, Transliteration,
 };
 use crate::sed::error_handling::{ScriptLocation, input_runtime_error};
 use crate::sed::fast_io::{IOChunk, LineReader, OutputBuffer};
@@ -40,30 +39,21 @@ macro_rules! extract_variant {
 /// Return true if the passed address matches the current I/O context.
 fn match_address(
     addr: &Address,
+    reader: &mut LineReader,
     pattern: &mut IOChunk,
     context: &mut ProcessingContext,
     location: &ScriptLocation,
 ) -> UResult<bool> {
-    match addr.atype {
+    match addr {
-        AddressType::Re => {
+        Address::Re(re) => {
-            if let AddressValue::Regex(ref re) = addr.value {
+            let regex = re_or_saved_re(re, context, location)?;
-                let regex = re_or_saved_re(re, context, location)?;
+            match regex.is_match(pattern) {
-                match regex.is_match(pattern) {
+                Ok(result) => Ok(result),
-                    Ok(result) => Ok(result),
+                Err(e) => input_runtime_error(location, context, e.to_string()),
-                    Err(e) => input_runtime_error(location, context, e.to_string()),
-                }
-            } else {
-                Ok(false)
             }
         }
 
-        AddressType::Line => {
+        Address::Line(lineno) => Ok(context.line_number == *lineno),
-            if let AddressValue::LineNumber(lineno) = addr.value {
-                Ok(context.line_number == lineno)
-            } else {
-                Ok(false)
-            }
-        }
 
         // Recognize "$" as the last line of last file. This is consistent
         // with the original 7th Research Edition implementation:
@@ -71,7 +61,7 @@ fn match_address(
         // The FreeBSD version checked for subsequent empty files, but this
         // can lead to destructive reads (e.g. from named pipes),
         // and is probably an overkill.
-        AddressType::Last => Ok(context.last_line && (context.last_file || context.separate)),
+        Address::Last => Ok(reader.last_line()? && (context.last_file || context.separate)),
 
         _ => panic!("invalid address type in match_address"),
     }
@@ -81,63 +71,66 @@ fn match_address(
 /// Return true if the command applies to the given pattern.
 fn applies(
     command: &mut Command,
+    reader: &mut LineReader,
     pattern: &mut IOChunk,
     context: &mut ProcessingContext,
 ) -> UResult<bool> {
     let linenum = context.line_number;
 
     let result = if command.addr1.is_none() && command.addr2.is_none() {
+        // No address
         Ok(true)
     } else if let Some(addr2) = &command.addr2 {
+        // Two addresses
         if let Some(start) = command.start_line {
-            match addr2.atype {
+            // Range is already latched active.
-                AddressType::RelLine => {
+            match addr2 {
-                    if let AddressValue::LineNumber(n) = addr2.value {
+                Address::RelLine(n) => {
-                        if linenum - start <= n {
+                    if linenum - start > *n {
-                            Ok(true)
+                        command.start_line = None;
-                        } else {
-                            command.start_line = None;
-                            Ok(false)
-                        }
-                    } else {
                         Ok(false)
+                    } else {
+                        Ok(true)
                     }
                 }
-                _ => {
+                Address::Line(n) => {
-                    if match_address(addr2, pattern, context, &command.location)? {
+                    // Special case: already ended
+                    if linenum > *n {
                         command.start_line = None;
-                        context.last_address = true;
+                        Ok(false)
-                        Ok(true)
-                    } else if addr2.atype == AddressType::Line {
-                        if let AddressValue::LineNumber(n) = addr2.value {
-                            if linenum > n {
-                                command.start_line = None;
-                                Ok(false)
-                            } else {
-                                Ok(true)
-                            }
-                        } else {
-                            Ok(true)
-                        }
                     } else {
                         Ok(true)
                     }
                 }
+                Address::StepMatch(step) => Ok((linenum - start).is_multiple_of(*step)),
+                Address::StepEnd(step) => {
+                    // Inclusive end on multiple of step
+                    if linenum.is_multiple_of(*step) {
+                        command.start_line = None;
+                    }
+                    Ok(true)
+                }
+                _ => {
+                    if match_address(addr2, reader, pattern, context, &command.location)? {
+                        command.start_line = None;
+                        context.last_address = true;
+                    }
+                    Ok(true)
+                }
             }
         } else if let Some(addr1) = &command.addr1 {
-            if match_address(addr1, pattern, context, &command.location)? {
+            // See if latch must start.
-                match addr2.atype {
+            if match_address(addr1, reader, pattern, context, &command.location)? {
-                    AddressType::Line => {
+                match addr2 {
-                        if let AddressValue::LineNumber(n) = addr2.value {
+                    Address::Line(n) => {
-                            if linenum >= n {
+                        if linenum >= *n {
-                                context.last_address = true;
+                            context.last_address = true;
-                            } else {
+                        } else {
-                                command.start_line = Some(linenum);
+                            command.start_line = Some(linenum);
-                            }
                         }
                     }
-                    AddressType::RelLine => {
+                    Address::RelLine(n) => {
-                        if let AddressValue::LineNumber(0) = addr2.value {
+                        if *n == 0 {
                             context.last_address = true;
                         } else {
                             command.start_line = Some(linenum);
@@ -155,9 +148,17 @@ fn applies(
             Ok(false)
         }
     } else if let Some(addr1) = &command.addr1 {
-        Ok(match_address(addr1, pattern, context, &command.location)?)
+        // Single address
+        Ok(match_address(
+            addr1,
+            reader,
+            pattern,
+            context,
+            &command.location,
+        )?)
     } else {
-        Ok(false)
+        // All allowed cases have been covered by the above logic.
+        panic!("impossible address combination");
     };
 
     if command.non_select {
@@ -431,8 +432,7 @@ fn process_file(
     context: &mut ProcessingContext,
 ) -> UResult<()> {
     // Loop over the input lines as pattern space.
-    'lines: while let Some((mut pattern, last_line)) = reader.get_line()? {
+    'lines: while let Some(mut pattern) = reader.get_line()? {
-        context.last_line = last_line;
         context.line_number += 1;
         context.substitution_made = false;
         // Set the script command from which to start.
@@ -455,7 +455,7 @@ fn process_file(
         while let Some(command_rc) = current.clone() {
             let mut command = command_rc.borrow_mut();
 
-            if !applies(&mut command, &mut pattern, context)? {
+            if !applies(&mut command, reader, &mut pattern, context)? {
                 // Advance to next command
                 current = command.next.clone();
                 continue;
@@ -494,7 +494,7 @@ fn process_file(
                     // At range end replace pattern space with text and
                     // start the next cycle.
                     pattern.clear();
-                    if command.addr2.is_none() || context.last_address || context.last_line {
+                    if command.addr2.is_none() || context.last_address || reader.last_line()? {
                         let text = extract_variant!(command, Text);
                         output.write_str(text.as_ref())?;
                     }
@@ -567,6 +567,10 @@ fn process_file(
                 'p' => {
                     // Write the pattern space to standard output.
                     write_chunk(output, context, &pattern)?;
+                    if !pattern.is_newline_terminated() {
+                        // !chomped equivalent
+                        output.write_str("\n")?; // explicit \n
+                    }
                 }
                 'P' => {
                     // Output pattern space, up to the first \n.
@@ -675,6 +679,21 @@ fn process_file(
     Ok(())
 }
 
+/// Mark all address ranges non-active (and 0-starting ones as active).
+fn reset_latched_address_ranges(range_commands: &mut [Rc<RefCell<Command>>]) {
+    for cmd_rc in range_commands.iter() {
+        let mut cmd = cmd_rc.borrow_mut();
+
+        cmd.start_line =
+            // Check for address-spec line 0 pre-latch extension.
+            if let Some(addr1) = &cmd.addr1 && matches!(addr1, Address::Line(0)) {
+                Some(0)
+            } else {
+                None
+            };
+    }
+}
+
 /// Process all input files
 pub fn process_all_files(
     commands: Option<Rc<RefCell<Command>>>,
@@ -692,9 +711,11 @@ pub fn process_all_files(
             .map_err_context(|| format!("error opening input file {}", path.quote()))?;
         let output = in_place.begin(path)?;
 
-        if context.separate {
+        if index == 0 || context.separate {
             context.line_number = 0;
+            reset_latched_address_ranges(&mut context.range_commands);
         }
+
         context.input_name = path.quote().to_string();
         process_file(&commands, &mut reader, output, context)?;
 

tests/by-util/test_sed.rs

@@ -14,6 +14,7 @@ use std::io::{Read, Write};
 #[cfg(unix)]
 use assert_fs::fixture::{FileWriteStr, PathChild};
 
+use sha2::{Digest, Sha256};
 use tempfile::NamedTempFile;
 use uutests::new_ucmd;
 
@@ -225,6 +226,15 @@ check_output!(addr_escaped_delimiter, ["-n", "\\_l1\\_7_p", LINES1]);
 check_output!(addr_range_numeric, ["-n", "1,4p", LINES1]);
 check_output!(addr_range_to_last, ["-n", "1,$p", LINES1, LINES2]);
 check_output!(addr_range_to_pattern, ["-n", "1,/l2_9/p", LINES1, LINES2]);
+check_output!(
+    addr_range_straddle,
+    ["-n", "/l1_3/,/l2_3/p", LINES1, LINES2]
+);
+check_output!(
+    addr_range_separate,
+    ["-n", "--separate", "/l1_3/,/l2_3/p", LINES1, LINES2]
+);
+check_output!(addr_range_from_zero_to_pattern, ["-n", "0,/_1/p", LINES1]);
 check_output!(addr_pattern_to_last, ["-n", "/4/,$p", LINES1, LINES2]);
 check_output!(addr_pattern_to_straddle, ["-n", "/4/,20p", LINES1, LINES2]);
 check_output!(addr_pattern_to_pattern, ["-n", "/4/,/10/p", LINES1, LINES2]);
@@ -259,6 +269,10 @@ check_output!(
 );
 check_output!(addr_empty_re_reuse, ["-n", "/_2/,//p", LINES1, LINES2]);
 check_output!(addr_simple_negation, ["-e", r"4,12!s/^/^/", LINES1]);
+check_output!(addr_range_even, ["-n", "0~2p", LINES1]);
+check_output!(addr_range_odd, ["-n", "1~2p", LINES1]);
+check_output!(addr_range_step_zero, ["-n", "10~0p", LINES1]);
+check_output!(addr_range_end_multiple, ["-n", "/l1_2/,~10p", LINES1]);
 
 ////////////////////////////////////////////////////////////
 // Substitution: s
@@ -987,6 +1001,42 @@ check_output!(pi, ["-f", "script/math.sed", "input/pi"]);
 // Solve the Towers of Hanoi puzzle
 check_output!(hanoi, ["-f", "script/hanoi.sed", "input/hanoi"]);
 
+////////////////////////////////////////////////////////////
+// Long-running scripts
+// Test with cargo test -- --ignored
+
+// Check the output of Bach's prelude in C major from WTC book I.
+// Run with cargo test test_bach_prelude_matches -- --ignored.
+#[test]
+#[ignore] // Slow; produces 5.8 MB of raw audio.
+fn test_bach_prelude_matches() {
+    let res = new_ucmd!()
+        .args(&["-E", "-f", "script/bach.sed"])
+        .pipe_in("\n")
+        .succeeds();
+
+    // Compare SHA-256 output against GNU sed output.
+    let mut hasher = Sha256::new();
+    hasher.update(res.stdout());
+    let digest = hasher.finalize();
+
+    let got = hex::encode(digest);
+    assert_eq!(
+        got,
+        "c4e50d6791a60692745e958dc48d43a40bccea2ce5cea31b7125a40604cc3219"
+    );
+}
+
+// Draw the Mandelbrot set.
+#[ignore] // Slow; takes > 15" on an i7 CPU
+#[test]
+fn test_mandelbrod() {
+    new_ucmd!()
+        .args(&["-En", "-f", "script/mandelbrot.sed", "input/newline"])
+        .succeeds()
+        .stdout_is_fixture("output/mandelbrot");
+}
+
 ////////////////////////////////////////////////////////////
 // Error handling
 #[test]
@@ -1016,6 +1066,51 @@ fn test_undefined_label() {
         .stderr_is("sed: <script argument 1>:1:1: error: undefined label `foo'\n");
 }
 
+#[test]
+fn test_addr0_non_posix() {
+    new_ucmd!()
+        .args(&["--posix", "0,/foo/p"])
+        .fails()
+        .code_is(1)
+        .stderr_is("sed: <script argument 1>:1:2: error: address 0 is invalid in POSIX mode\n");
+}
+
+#[test]
+fn test_addr0_second_required() {
+    new_ucmd!()
+        .args(&["0p"])
+        .fails()
+        .code_is(1)
+        .stderr_is("sed: <script argument 1>:1:2: error: address 0 requires a second address\n");
+}
+
+#[test]
+fn test_addr0_second_re_only() {
+    new_ucmd!()
+        .args(&["0,4p"])
+        .fails()
+        .code_is(1)
+        .stderr_is("sed: <script argument 1>:1:4: error: address 0 can only be used with a regular expression or ~step\n");
+}
+
+#[test]
+fn test_step_match_non_posix() {
+    new_ucmd!()
+        .args(&["--posix", "3~2p"])
+        .fails()
+        .code_is(1)
+        .stderr_is("sed: <script argument 1>:1:3: error: ~step is invalid in POSIX mode\n");
+}
+
+#[test]
+fn test_step_end_non_posix() {
+    new_ucmd!()
+        .args(&["--posix", "3,~2p"])
+        .fails()
+        .code_is(1)
+        .stderr_is("sed: <script argument 1>:1:4: error: ~step is invalid in POSIX mode\n");
+}
+
 // The following test diverse ways in which regexes are matched.
 // Search for 'regex\.' to find them in the code.
 #[test]
@@ -1080,3 +1175,14 @@ fn test_missing_address_re() {
         .code_is(2)
         .stderr_is("sed: <script argument 1>:2:3: 'input/lines1':1 error: no previous regular expression\n");
 }
+
+////////////////////////////////////////////////////////////
+// Test for issue #143: Missing newline in output with `-e p`
+#[test]
+fn test_print_command_adds_newline() {
+    new_ucmd!()
+        .args(&["-e", "p"])
+        .pipe_in("foo")
+        .succeeds()
+        .stdout_is("foo\nfoo");
+}

tests/fixtures/sed/input/newline

@@ -0,0 +1 @@
+

tests/fixtures/sed/output/addr_range_end_multiple

@@ -0,0 +1,9 @@
+l1_2
+l1_3
+l1_4
+l1_5
+l1_6
+l1_7
+l1_8
+l1_9
+l1_10

tests/fixtures/sed/output/addr_range_even

@@ -0,0 +1,7 @@
+l1_2
+l1_4
+l1_6
+l1_8
+l1_10
+l1_12
+l1_14

tests/fixtures/sed/output/addr_range_from_zero_to_pattern

@@ -0,0 +1 @@
+l1_1