NLPPlus 2.0.20__tar.gz → 2.1.0__tar.gz

{nlpplus-2.0.20 → nlpplus-2.1.0}/NLPPlus/_version.py

@@ -18,7 +18,7 @@ version_tuple: tuple[int | str, ...]
 commit_id: str | None
 __commit_id__: str | None
 
-__version__ = version = '2.0.20'
-__version_tuple__ = version_tuple = (2, 0, 20)
+__version__ = version = '2.1.0'
+__version_tuple__ = version_tuple = (2, 1, 0)
 
-__commit_id__ = commit_id = 'gceab88c8c'
+__commit_id__ = commit_id = 'ge70676e9c'

{nlpplus-2.0.20 → nlpplus-2.1.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.2
 Name: NLPPlus
-Version: 2.0.20
+Version: 2.1.0
 Summary: Python extension for the NLP++ text analysis engine
 Keywords: nlp,text
 Author-Email: David De Hilster <contact@visualtext.org>

{nlpplus-2.0.20 → nlpplus-2.1.0}/nlp-engine/.github/workflows/build-enginefiles.yml

@@ -4,10 +4,6 @@ on:
   workflow_dispatch:
   pull_request:
     types: [opened, synchronize, reopened]
-  # Build the tagged commit so release assets match the tagged version.
-  push:
-    tags:
-      - '*'
 
 jobs:
   zip_files:

{nlpplus-2.0.20 → nlpplus-2.1.0}/nlp-engine/.github/workflows/build-linux.yml

@@ -4,10 +4,6 @@ on:
   workflow_dispatch:
   pull_request:
     types: [opened, synchronize, reopened]
-  # Build the tagged commit so the release binary carries the right version.
-  push:
-    tags:
-      - '*'
 
 jobs:
   build_engine:
@@ -46,27 +42,54 @@ jobs:
       # ===============================
       - name: Build and Run (Ubuntu 20.04)
         if: matrix.id == 'docker-ubuntu-20.04'
-        shell: bash
-        run: |
-          set -euo pipefail
-          # Pull the target image directly (with retries) and run the build
-          # script in it via the runner's pre-installed Docker. This replaces
-          # addnab/docker-run-action@v3, which first builds a docker:20.10
-          # helper image -- that Docker Hub pull was intermittently timing out
-          # ("failed to resolve docker.io/library/docker:20.10 ... i/o timeout")
-          # and failing the whole Linux job.
-          for attempt in 1 2 3 4 5; do
-            if docker pull ubuntu:20.04; then break; fi
-            echo "docker pull ubuntu:20.04 failed (attempt $attempt); retrying in 20s..."
-            sleep 20
-          done
-          docker run --rm \
-            --shm-size=512m \
-            --mount type=tmpfs,destination=/dev/shm \
-            -v "${{ github.workspace }}:/workspace" \
-            -w /workspace \
-            ubuntu:20.04 \
-            bash scripts/build-ubuntu-2004.sh
+        uses: addnab/docker-run-action@v3
+        with:
+          image: ubuntu:20.04
+          options: >-
+            --shm-size=512m
+            --mount type=tmpfs,destination=/dev/shm
+            -v ${{ github.workspace }}:/workspace
+            -w /workspace
+          run: |
+            set -e
+            export DEBIAN_FRONTEND=noninteractive
+            export TZ=Etc/UTC
+            apt update
+            apt install -y \
+              git build-essential cmake g++ curl unzip zip tar \
+              autoconf-archive pkg-config python3 ninja-build libicu66 libicu-dev
+            rm -rf analyzers   # drop the pinned submodule copy; test wants fresh analyzers master
+            git clone --recurse-submodules https://github.com/VisualText/analyzers.git analyzers
+            mkdir -p build
+            cd build
+            cmake ..
+            make -j$(nproc)
+            cd ..
+            mkdir -p rfb-logs data/rfb/spec analyzers/parse-en-us/output
+            cp ./bin/nlp ./bin/nlpl.exe
+            echo "Copying ICU 66 libraries..."
+            mkdir -p icu-libs
+            cp /usr/lib/x86_64-linux-gnu/libicu*.so.66* ./icu-libs/
+            zip -r icu-libs.zip icu-libs
+            ./bin/nlp --version || true
+            ./bin/nlp -ANA ./analyzers/parse-en-us -WORK ./ ./analyzers/parse-en-us/input/doj.txt -LOG rfb-logs -DEV || true
+
+            echo "Packaging compile libraries..."
+            rm -rf compile-libs
+            mkdir -p compile-libs/include/Api compile-libs/lib
+            cp -r include/Api/. compile-libs/include/Api/
+            cp -r cs/include compile-libs/include/cs
+            for name in prim kbm consh words lite; do
+              lib=$(find . -type f -name "lib${name}.a" -not -path "./vcpkg/*" -not -path "./compile-libs/*" | head -n1)
+              if [ -z "$lib" ]; then
+                echo "Missing required static library: lib${name}.a" >&2
+                exit 1
+              fi
+              cp "$lib" "compile-libs/lib/lib${name}.a"
+            done
+            # System ICU 66 runtime libs (dynamic) used by the 20.04 build.
+            cp /usr/lib/x86_64-linux-gnu/libicu*.so.66* compile-libs/lib/
+            (cd compile-libs && zip -r ../nlpengine-compile-libs-linux-20.04.zip .)
 
       - name: Upload ICU Libraries Zip (20.04)
         if: matrix.id == 'docker-ubuntu-20.04'
@@ -122,19 +145,6 @@ jobs:
         if: matrix.id != 'docker-ubuntu-20.04'
         run: ./bin/nlp --version || true
 
-      - name: Full-dict binary-search hang regression (#669, Non-20.04)
-        if: matrix.id != 'docker-ubuntu-20.04'
-        shell: bash
-        run: |
-          # An em-dash sorts after every word in the lazy "*full.dict"; that used
-          # to make the dict binary search loop forever (#669: afterKey=-1 at EOF
-          # -> lo=-1). The fixture is self-contained (tiny sorted zz-full.dict +
-          # em-dash input). A hang -> `timeout` exits 124 -> this step fails.
-          # NOTE: no `|| true` here on purpose -- we WANT a hang to fail CI.
-          timeout 120 ./bin/nlp -ANA .github/workflows/tests/fulldict-hang \
-            -WORK ./ .github/workflows/tests/fulldict-hang/input/dash.txt
-          echo "[#669 regression] analyzer completed without hanging"
-
       - name: Copy nlp.exe to nlpl.exe (Non-20.04)
         if: matrix.id != 'docker-ubuntu-20.04'
         run: cp bin/nlp bin/nlpl.exe

{nlpplus-2.0.20 → nlpplus-2.1.0}/nlp-engine/.github/workflows/build-macos.yml

@@ -4,11 +4,7 @@ on:
   workflow_dispatch:
   pull_request:
     types: [opened, synchronize, reopened]
-  # Build the tagged commit so the release binary carries the right version.
-  push:
-    tags:
-      - '*'
-
+  
 jobs:
   build_engine:
     name: Build NLP-ENGINE

{nlpplus-2.0.20 → nlpplus-2.1.0}/nlp-engine/.github/workflows/build-visualfiles.yml

@@ -4,12 +4,8 @@ on:
   workflow_dispatch:
   pull_request:
     types: [opened, synchronize, reopened]
-  # Build the tagged commit so release assets match the tagged version.
-  push:
-    tags:
-      - '*'
-
-jobs:
+    
+jobs:    
   zip_files:
     runs-on: ubuntu-latest
     steps:

{nlpplus-2.0.20 → nlpplus-2.1.0}/nlp-engine/.github/workflows/build-windows.yml

@@ -6,10 +6,6 @@ on:
     types: [opened, synchronize, reopened]
     branches:
       - master
-  # Build the tagged commit itself so the release binary carries the right
-  # version. (A `tags:` filter under `pull_request` is ignored by GitHub --
-  # it must live under `push:`.)
-  push:
     tags:
       - '*'
 
@@ -49,18 +45,6 @@ jobs:
       - name: Test executable --version
         run: ./bin/Release/nlp.exe --version || true
 
-      - name: Full-dict binary-search hang regression (#669)
-        # An em-dash sorts after every word in the lazy "*full.dict"; pre-fix the
-        # dict binary search looped forever (afterKey=-1 at EOF -> lo=-1). The
-        # self-contained fixture must complete; if it hangs, this step exceeds
-        # timeout-minutes and fails the job (no `|| true`).
-        timeout-minutes: 3
-        shell: bash
-        run: |
-          ./bin/Release/nlp.exe -ANA .github/workflows/tests/fulldict-hang \
-            -WORK ./ .github/workflows/tests/fulldict-hang/input/dash.txt
-          echo "[#669 regression] analyzer completed without hanging"
-
       - name: Copy nlp.exe to nlpw.exe
         run: cp bin/Release/nlp.exe bin/Release/nlpw.exe
         shell: bash

{nlpplus-2.0.20 → nlpplus-2.1.0}/nlp-engine/.github/workflows/move-assets.yml

@@ -5,25 +5,15 @@ permissions:
   actions: write
 
 on:
-  # Manual re-release: pass an explicit tag (e.g. v3.5.4) to rebuild that
-  # release's assets from the exact tagged commit.
   workflow_dispatch:
-    inputs:
-      tag:
-        description: "Tag to (re)build release assets for, e.g. v3.5.4. Defaults to the latest tag."
-        required: false
-        type: string
-  # Primary trigger: a tag push. We then WAIT for that tag's builds to finish
-  # (see "Find workflow run IDs") rather than grabbing whatever build is newest.
+  workflow_run:
+    workflows: ["build-windows.yml", "build-macos.yml", "build-enginefiles.yml", "build-visualfiles.yml", "build-linux.yml"]
+    types:
+      - completed
   push:
     tags:
     - '*'
 
-# Never let two asset-moves touch the same release at once.
-concurrency:
-  group: move-assets-${{ github.event.inputs.tag || github.ref }}
-  cancel-in-progress: false
-
 jobs:
   download_assets:
     runs-on: ubuntu-latest
@@ -35,104 +25,56 @@ jobs:
         fetch-depth: 0  # Fetch all history and tags
 
     
-    - name: Extract tag name and resolve its commit
+    - name: Extract tag name from ref or get latest tag
       id: get_tag
       run: |
-        git fetch --tags --force
-        if [[ -n "${{ github.event.inputs.tag }}" ]]; then
-          # Manual re-release for a specific tag.
-          TAG_NAME="${{ github.event.inputs.tag }}"
-          echo "Using tag from workflow_dispatch input: $TAG_NAME"
-        elif [[ "${{ github.ref }}" == refs/tags/* ]]; then
-          # Triggered by a tag push.
+        if [[ "${{ github.ref }}" == refs/tags/* ]]; then
+          # When directly triggered by a tag push
           TAG_NAME=${GITHUB_REF#refs/tags/}
           echo "Using tag from push trigger: $TAG_NAME"
         else
-          # Fallback: latest tag.
+          # When triggered by workflow_run or workflow_dispatch
+          # Get the latest tag from the repository
+          git fetch --tags
           TAG_NAME=$(git describe --tags --abbrev=0)
-          echo "Using latest tag from repository: $TAG_NAME"
-        fi
-
-        if [ -z "$TAG_NAME" ]; then
-          echo "ERROR: could not determine a tag to release." >&2
-          exit 1
+          
+          # If no tag found, use timestamp
+          if [ -z "$TAG_NAME" ]; then
+            TIMESTAMP=$(date +'%Y.%m.%d-%H%M')
+            TAG_NAME="v${TIMESTAMP}"
+            echo "No tag found, using generated version: $TAG_NAME"
+          else
+            echo "Using latest tag from repository: $TAG_NAME"
+          fi
         fi
-
-        # The commit this tag points to. The release MUST attach binaries built
-        # from THIS commit, not whatever build happened to be newest -- that bug
-        # stapled a 3.5.3 binary onto the v3.5.4 release.
-        TAG_SHA=$(git rev-list -n1 "$TAG_NAME")
+        
+        # Store the tag name for later use
         echo "TAG_NAME=$TAG_NAME" >> $GITHUB_ENV
-        echo "TAG_SHA=$TAG_SHA" >> $GITHUB_ENV
-        echo "Selected TAG_NAME=$TAG_NAME -> TAG_SHA=$TAG_SHA"
-
-    - name: Hide any empty pre-existing release during the build wait
-      env:
-        GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
-      run: |
-        set -euo pipefail
-        # If a release for this tag was created up front (e.g. via the GitHub UI)
-        # it is published but empty until the builds finish ~15 min later. An
-        # update-checker hitting /releases/latest (the VS Code NLP++ extension
-        # does this) would see a new version with no downloadable assets and
-        # break for that window. Mark such a release a draft now; the "Create
-        # Release" step below republishes it with assets (draft: false). If a
-        # build fails, it safely stays a draft instead of shipping empty.
-        rel=$(gh api "repos/${{ github.repository }}/releases/tags/$TAG_NAME" 2>/dev/null || echo '{}')
-        is_draft=$(echo "$rel" | jq -r '.draft // empty')
-        n_assets=$(echo "$rel" | jq -r '(.assets // []) | length')
-        if [ "$is_draft" = "false" ] && [ "${n_assets:-0}" = "0" ]; then
-          echo "Release $TAG_NAME is published but empty -> drafting it until assets are ready."
-          gh release edit "$TAG_NAME" --repo "${{ github.repository }}" --draft
-        else
-          echo "Nothing to hide (exists=$([ -n "$is_draft" ] && echo yes || echo no), draft=$is_draft, assets=${n_assets:-0})."
-        fi
-
-    - name: Find workflow run IDs for the tagged commit
-      env:
-        GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+        
+        # Debug output
+        echo "Selected TAG_NAME: $TAG_NAME"
+    - name: Find workflow run IDs
       run: |
-        set -euo pipefail
-        # Pick the build run whose head_sha == this tag's commit, and WAIT for it
-        # to finish. Previously this grabbed the newest *successful* build of any
-        # branch/PR, which stapled a stale (wrong-version) binary onto the
-        # release. Each build workflow now triggers on the tag push, so a run for
-        # $TAG_SHA exists; we just have to wait for it to go green.
-        wait_for_run() {
-          local wf="$1"
-          local timeout=5400 interval=30 waited=0   # up to 90 min
-          while true; do
-            local json id pending bad
-            json=$(gh api "repos/${{ github.repository }}/actions/workflows/$wf/runs?head_sha=$TAG_SHA&per_page=100" 2>/dev/null || echo '{}')
-            id=$(echo "$json" | jq -r '(.workflow_runs // []) | map(select(.conclusion=="success")) | .[0].id // empty')
-            if [ -n "$id" ]; then echo "$id"; return 0; fi
-            pending=$(echo "$json" | jq -r '(.workflow_runs // []) | map(select(.status!="completed")) | length')
-            bad=$(echo "$json" | jq -r '(.workflow_runs // []) | map(select(.conclusion=="failure" or .conclusion=="cancelled" or .conclusion=="timed_out" or .conclusion=="startup_failure")) | .[0].id // empty')
-            if [ "$pending" = "0" ] && [ -n "$bad" ]; then
-              echo "ERROR: $wf for $TAG_SHA finished without success (run $bad)." >&2
-              return 1
-            fi
-            if [ "$waited" -ge "$timeout" ]; then
-              echo "ERROR: timed out after ${timeout}s waiting for a successful $wf run for $TAG_SHA." >&2
-              return 1
-            fi
-            echo "Waiting for $wf build of $TAG_SHA to succeed... (${waited}s elapsed)" >&2
-            sleep "$interval"; waited=$((waited+interval))
-          done
-        }
-
-        WIN_RUN_ID=$(wait_for_run build-windows.yml)
+        echo "Finding most recent successful workflow runs..."
+        
+        WIN_RUN_ID=$(gh api repos/${{ github.repository }}/actions/workflows/build-windows.yml/runs --paginate | jq -r ".workflow_runs[] | select(.conclusion == \"success\") | .id" | head -n 1)
+        echo "Windows run ID: $WIN_RUN_ID"
         echo "WIN_RUN_ID=$WIN_RUN_ID" >> $GITHUB_ENV
-        LINUX_RUN_ID=$(wait_for_run build-linux.yml)
+    
+        LINUX_RUN_ID=$(gh api repos/${{ github.repository }}/actions/workflows/build-linux.yml/runs --paginate | jq -r ".workflow_runs[] | select(.conclusion == \"success\") | .id" | head -n 1)
+        echo "Linux run ID: $LINUX_RUN_ID"
         echo "LINUX_RUN_ID=$LINUX_RUN_ID" >> $GITHUB_ENV
-        MAC_RUN_ID=$(wait_for_run build-macos.yml)
+        
+        MAC_RUN_ID=$(gh api repos/${{ github.repository }}/actions/workflows/build-macos.yml/runs --paginate | jq -r ".workflow_runs[] | select(.conclusion == \"success\") | .id" | head -n 1)
         echo "MAC_RUN_ID=$MAC_RUN_ID" >> $GITHUB_ENV
-        ENGINE_RUN_ID=$(wait_for_run build-enginefiles.yml)
+    
+        ENGINE_RUN_ID=$(gh api repos/${{ github.repository }}/actions/workflows/build-enginefiles.yml/runs --paginate | jq -r ".workflow_runs[] | select(.conclusion == \"success\") | .id" | head -n 1)
         echo "ENGINE_RUN_ID=$ENGINE_RUN_ID" >> $GITHUB_ENV
-        VISUAL_RUN_ID=$(wait_for_run build-visualfiles.yml)
+    
+        VISUAL_RUN_ID=$(gh api repos/${{ github.repository }}/actions/workflows/build-visualfiles.yml/runs --paginate | jq -r ".workflow_runs[] | select(.conclusion == \"success\") | .id" | head -n 1)
         echo "VISUAL_RUN_ID=$VISUAL_RUN_ID" >> $GITHUB_ENV
-
-        echo "Resolved runs for $TAG_SHA: win=$WIN_RUN_ID linux=$LINUX_RUN_ID mac=$MAC_RUN_ID engine=$ENGINE_RUN_ID visual=$VISUAL_RUN_ID"
+      env:
+        GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
 
     # Windows artifacts with specific run ID
     - name: Fetch Windows exe
@@ -333,14 +275,9 @@ jobs:
       with:
         tag_name: ${{ env.TAG_NAME }}
         name: "Release ${{ env.TAG_NAME }}"
-        # Publish (un-draft) now that the assets below are attached. This also
-        # re-publishes a release the draft-guard step hid at the start.
-        draft: false
         body: |
           NLP Engine Release ${{ env.TAG_NAME }}
-
-          Built from commit ${{ env.TAG_SHA }}
-
+        
           Build information:
           - Windows build: ${{ env.WIN_RUN_ID }}
           - Linux build: ${{ env.LINUX_RUN_ID }}
@@ -366,11 +303,7 @@ jobs:
       env:
         GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
 
-    # Percolate downstream only on a genuine new release (tag push). A manual
-    # re-release (workflow_dispatch, e.g. re-attaching an old tag's binary) must
-    # NOT re-bump downstream packages to that older engine version.
     - name: Trigger nlp-engine-windows workflow
-      if: github.event_name == 'push'
       uses: peter-evans/repository-dispatch@v2
       with:
         token: ${{ secrets.CLASSIC_PAT }}
@@ -380,7 +313,6 @@ jobs:
       continue-on-error: true
 
     - name: Trigger nlp-engine-mac workflow
-      if: github.event_name == 'push'
       uses: peter-evans/repository-dispatch@v2
       with:
         token: ${{ secrets.CLASSIC_PAT }}
@@ -390,34 +322,10 @@ jobs:
       continue-on-error: true
 
     - name: Trigger nlp-engine-linux workflow
-      if: github.event_name == 'push'
       uses: peter-evans/repository-dispatch@v2
       with:
         token: ${{ secrets.CLASSIC_PAT }}
         repository: VisualText/nlp-engine-linux
         event-type: nlp-engine-release
         client-payload: '{"tag_name": "${{ env.TAG_NAME }}"}'
-      continue-on-error: true
-
-    # Package repos build the engine from source via their nlp-engine submodule,
-    # so an engine release must bump that submodule + republish. See
-    # docs/PERCOLATION.md ("nlp-engine -> npm/py packages").
-    - name: Trigger py-package-nlpengine workflow
-      if: github.event_name == 'push'
-      uses: peter-evans/repository-dispatch@v2
-      with:
-        token: ${{ secrets.CLASSIC_PAT }}
-        repository: VisualText/py-package-nlpengine
-        event-type: nlp-engine-release
-        client-payload: '{"tag_name": "${{ env.TAG_NAME }}"}'
-      continue-on-error: true
-
-    - name: Trigger npm-package-nlpengine workflow
-      if: github.event_name == 'push'
-      uses: peter-evans/repository-dispatch@v2
-      with:
-        token: ${{ secrets.CLASSIC_PAT }}
-        repository: VisualText/npm-package-nlpengine
-        event-type: nlp-engine-release
-        client-payload: '{"tag_name": "${{ env.TAG_NAME }}"}'
-      continue-on-error: true
+      continue-on-error: true  

{nlpplus-2.0.20 → nlpplus-2.1.0}/nlp-engine/cs/libconsh/cg.cpp

@@ -4201,13 +4201,6 @@ bool CG::nextKBBKey(FullFile &f, std::streamoff &keyStart, std::string &key, std
 			keyStart = pos;
 			key = w;
 			afterKey = f.stream.tellg();
-			// Reading the file's last line (no trailing newline) leaves the
-			// stream at EOF, where tellg() returns -1. The position past the
-			// last key is end-of-file; returning -1 made searchKBBFile set
-			// lo = -1, so the binary search never terminated for any key that
-			// sorts after the last word in the file (e.g. an em-dash). See #669.
-			if (afterKey < 0)
-				afterKey = f.size;
 			return true;
 		}
 	}
@@ -4237,13 +4230,6 @@ bool CG::nextDictKey(FullFile &f, std::streamoff &keyStart, std::string &key, st
 		keyStart = pos;
 		key = w;
 		afterKey = f.stream.tellg();
-		// Reading the file's last line (no trailing newline) leaves the stream
-		// at EOF, where tellg() returns -1. The position past the last key is
-		// end-of-file; returning -1 made searchDictFile set lo = -1, so the
-		// binary search never terminated for any key that sorts after the last
-		// word in the file (e.g. an em-dash U+2014 / en-dash U+2013). See #669.
-		if (afterKey < 0)
-			afterKey = f.size;
 		return true;
 	}
 }
@@ -4403,11 +4389,9 @@ CONCEPT *CG::searchKBBFile(FullFile &f, _TCHAR *str)
 			hi = mid;
 			continue;
 		}
-		if (strcmp(key.c_str(), str) < 0) {
-			if (afterKey <= lo)		// Defensive: never move lo backwards/sideways
-				break;				// (would loop forever). str is past EOF here.
+		if (strcmp(key.c_str(), str) < 0)
 			lo = afterKey;
-		} else
+		else
 			hi = mid;
 	}
 
@@ -4486,11 +4470,9 @@ CONCEPT *CG::searchDictFile(FullFile &f, _TCHAR *str)
 			hi = mid;
 			continue;
 		}
-		if (strcmp(key.c_str(), str) < 0) {
-			if (afterKey <= lo)		// Defensive: never move lo backwards/sideways
-				break;				// (would loop forever). str is past EOF here.
+		if (strcmp(key.c_str(), str) < 0)
 			lo = afterKey;
-		} else
+		else
 			hi = mid;
 	}
 

{nlpplus-2.0.20 → nlpplus-2.1.0}/nlp-engine/cs/libconsh/consh_kb.cpp

@@ -71,10 +71,7 @@ bool dirt;	// 06/29/03 AM.
 
 if (cg->cg_CONCEPT)
    {
-   // Root already exists: re-issuing ADD ROOT on an initialized KB is an
-   // expected, harmless no-op (every kb/dict and lazy "*-full" load re-runs
-   // it), so skip silently instead of spamming stderr. Uncomment to trace.
-   // std::_t_cerr << _T("[con_add_root: Root concept of kb already exists.]") << std::endl;
+   std::_t_cerr << _T("[con_add_root: Root concept of kb already exists.]") << std::endl;
    return;
    }
 

{nlpplus-2.0.20 → nlpplus-2.1.0}/nlp-engine/cs/libqconsh/consh_kb.cpp

@@ -76,10 +76,7 @@ XCON_S *con = &con_x;	// 02/14/07 AM.
 
 if (cg->cg_CONCEPT)
    {
-   // Root already exists: re-issuing ADD ROOT on an initialized KB is an
-   // expected, harmless no-op (every kb/dict and lazy "*-full" load re-runs
-   // it), so skip silently instead of spamming stderr. Uncomment to trace.
-   // _t_cerr << _T("[con_add_root: Root concept of kb already exists.]") << endl;
+   _t_cerr << _T("[con_add_root: Root concept of kb already exists.]") << endl;
    return;
    }
 

{nlpplus-2.0.20 → nlpplus-2.1.0}/nlp-engine/data/rfb/spec/xvar.nlp

@@ -1,21 +1,16 @@
 ###############################################
 # FILE:     XVAR.PAT
-# SUBJ:     Merge _xVAR("attr") / _xVAR(attr) into a single nonliteral token.
+# SUBJ:     Merge _xVAR("attr") into a single nonliteral match-list token.
 # NOTE:     Runs after bigtok forms _NONLIT/_STR and whitespace is zapped,
 #           and before list collects parenthesized match lists, so the inner
-#           parens of _xVAR(attr) never reach the list grammar.  The merged
+#           parens of _xVAR("attr") never reach the list grammar.  The merged
 #           token _xVAR("attr") then flows through as one restricted-wildcard
 #           match-list entry; Pat::modeMatch1 tests the node attribute at
-#           runtime.  The attribute may be quoted (_STR) or an unquoted bareword
-#           (_LIT/_NUM); Pat::xvarMatch strips the optional quotes either way.
+#           runtime.
 ###############################################
 
 @POST
 	rfaxvar(1,3)
 	single()
 @RULES
-# Quoted attr:    _xVAR("attr")
 _NONLIT [base] <- _NONLIT \( _STR \) @@
-# Unquoted attr:  _xVAR(attr)  /  _xVAR(123)
-_NONLIT [base] <- _NONLIT \( _LIT \) @@
-_NONLIT [base] <- _NONLIT \( _NUM \) @@

{nlpplus-2.0.20 → nlpplus-2.1.0}/nlp-engine/docs/PERCOLATION.md

@@ -28,11 +28,7 @@ analyzers ──(after its auto-bump release)──► analyzers-release ──
 
 nlp-engine ──(release / move-assets)──► nlp-engine-release ──► nlp-engine-linux
                                                            ├──► nlp-engine-windows
-                                                           ├──► nlp-engine-mac
-                                                           ├──► npm-package-nlpengine
-                                                           └──► py-package-nlpengine
-                                                                (each bumps the nlp-engine
-                                                                 submodule + tags → publishes)
+                                                           └──► nlp-engine-mac
 ```
 
 ## Mechanism
@@ -61,7 +57,7 @@ which has a workflow triggered by `repository_dispatch: types: [<event>]`.
 | package-analyzers | `v*` tag (auto-created by `tag-on-push.yml` on any analyzer edit pushed to main, or by `update-parse-en-us.yml` on a parse-en-us release) | `package-analyzers-release` | npm-package-nlpengine, py-package-nlpengine | `dispatch-update-package-analyzers.yml` |
 | analyzer-templates | `v*` tag | `analyzer-templates-release` | nlp-engine-{linux,windows,mac}, visualtext-files | `dispatch-update-analyzer-templates.yml` |
 | analyzers | end of its bump job | `analyzers-release` | nlp-engine | `parse-en-us.yml` (final step) |
-| nlp-engine | `v*` tag push | `nlp-engine-release` | nlp-engine-{linux,windows,mac}, npm-package-nlpengine, py-package-nlpengine | `move-assets.yml` |
+| nlp-engine | release / build complete | `nlp-engine-release` | nlp-engine-{linux,windows,mac} | `move-assets.yml` |
 
 ## Who listens for what
 
@@ -77,8 +73,6 @@ which has a workflow triggered by `repository_dispatch: types: [<event>]`.
 | nlp-engine | `analyzers-release` | `update-analyzers.yml` | open a PR bumping the analyzers submodule (no auto-release) |
 | nlp-engine-{linux,windows,mac} | `analyzer-templates-release` | `update-analyzer-templates.yml` | bump analyzer-templates submodule |
 | nlp-engine-{linux,windows,mac} | `nlp-engine-release` | `nlp-engine-build.yml` | build the platform package |
-| npm-package-nlpengine | `nlp-engine-release` | `update-nlp-engine.yml` | bump the `nlp-engine` submodule, `npm version patch`, push v* tag → `publish.yml` publishes to npm |
-| py-package-nlpengine | `nlp-engine-release` | `update-nlp-engine.yml` | bump the `nlp-engine` submodule, push next v* tag → `publish.yml` publishes to PyPI |
 
 ## Submodule embedding (what bumps what)
 
@@ -115,9 +109,6 @@ which has a workflow triggered by `repository_dispatch: types: [<event>]`.
   tag fires each package's `publish.yml`, which builds and publishes to npm
   (OIDC trusted publisher) / PyPI (trusted publishing). No human step — editing an
   analyzer in `package-analyzers` lands new releases on both registries.
-  Likewise, on a `nlp-engine-release` ping, `update-nlp-engine.yml` bumps the
-  `nlp-engine` submodule and republishes — so a tagged **engine** release also
-  lands on both registries with no manual submodule bump.
   - ⚠️ The tag push **must** use `CLASSIC_PAT`; a `GITHUB_TOKEN`-pushed tag does
     not trigger `publish.yml`. So `CLASSIC_PAT` is required as a secret in
     `npm-package-nlpengine` and `py-package-nlpengine` (not just the senders).
@@ -140,32 +131,6 @@ its submodule pointer is already current.
    version bump to `patch` on a ping (a `repository_dispatch` event has no
    `inputs.bump`).
 
-## Engine release binaries: attach the build of the *tagged* commit
-
-The engine's per-platform binaries (`nlpw.exe`, `nlpm.exe`, Linux zips, ICU/compile
-libs) are built by `build-windows.yml` / `build-macos.yml` / `build-linux.yml` /
-`build-enginefiles.yml` / `build-visualfiles.yml`, then collected onto the GitHub
-Release by `move-assets.yml`.
-
-Two rules keep the attached binary's version in sync with the tag:
-
-1. **Each build workflow triggers on the tag push** (`push: tags: ['*']`), so the
-   tagged commit itself is compiled. (`NLP_ENGINE_VERSION` is baked into the binary
-   at compile time, so the binary must be built *from the release commit*.)
-2. **`move-assets.yml` selects the build run whose `head_sha` == the tag's commit**
-   and *waits* for it to go green, instead of grabbing whatever build is newest.
-
-> **History:** these workflows originally only built on `pull_request` (the
-> `build-windows.yml` `tags:` filter was mis-nested under `pull_request`, where
-> GitHub ignores it), and `move-assets` grabbed the *most recent successful* build
-> of any branch. The result: v3.5.4 shipped a binary from an older PR whose
-> `main.cpp` still said 3.5.3. Tag-SHA pinning + tag-triggered builds fix this.
-
-To **re-release a tag** (e.g. fix a bad asset): Actions → "Move Assets to Release"
-→ Run workflow, and set the **`tag`** input (e.g. `v3.5.4`). It resolves that tag's
-commit, waits for that commit's builds, and re-attaches. If those builds don't
-exist yet, dispatch each `build-*` workflow on the tag ref first.
-
 ## Gotcha: re-attaching a submodule that workflows fetch manually
 
 `nlp-engine` build/test jobs historically `git clone`d `analyzers` fresh into

{nlpplus-2.0.20 → nlpplus-2.1.0}/nlp-engine/include/Api/lite/nlp_engine.h

@@ -148,8 +148,4 @@ public:
 private:
     int createDir(_TCHAR *dirPath);
     int readFiles(_TCHAR *directoryPath);
-    // Canonical registry name for an analyzer (its path basename). init() and
-    // close() must key the VTRun analyzer registry on this so the same analyzer
-    // passed as different path-forms maps to one entry -- see #658 / #380.
-    void anaRegName(const _TCHAR *path, _TCHAR *out);
 };