NLPPlus 2.0.3__tar.gz → 2.0.5__tar.gz

{nlpplus-2.0.3 → nlpplus-2.0.5}/.github/workflows/publish.yml

@@ -17,7 +17,7 @@ jobs:
     steps:
       - uses: actions/checkout@v4
         with:
-          submodules: true
+          submodules: recursive
 
       - name: Build wheels
         uses: pypa/cibuildwheel@v2.19.2
@@ -44,7 +44,7 @@ jobs:
     steps:
       - uses: actions/checkout@v4
         with:
-          submodules: true
+          submodules: recursive
 
       - name: Install packages
         run: brew install autoconf-archive autoconf automake pkg-config
@@ -72,7 +72,7 @@ jobs:
     steps:
       - uses: actions/checkout@v4
         with:
-          submodules: true
+          submodules: recursive
 
       - name: vcpkg build
         id: vcpkg

{nlpplus-2.0.3 → nlpplus-2.0.5}/.github/workflows/test.yml

@@ -12,7 +12,7 @@ jobs:
     steps:
     - uses: actions/checkout@v4
       with:
-        submodules: true
+        submodules: recursive
     - uses: actions/setup-python@v4
       with:
         python-version: '3.10'

nlpplus-2.0.5/.github/workflows/update-analyzers.yml

@@ -0,0 +1,89 @@
+# Install at: VisualText/py-package-nlpengine -> .github/workflows/update-analyzers.yml
+#
+# Listener in the cross-repo percolation chain (see
+# VisualText/nlp-engine/docs/PERCOLATION.md).
+#
+# Fully automatic: when package-analyzers releases (event-type
+# package-analyzers-release), bump the `NLPPlus/analyzers` submodule, commit, and
+# push the next v* tag. The tag fires publish.yml, which builds the wheels +
+# sdist and publishes to PyPI (trusted publishing). The package version is
+# derived from the tag by setuptools_scm, so no file edit is needed.
+#
+# The commit + tag are pushed with CLASSIC_PAT so the tag actually triggers
+# publish.yml (a GITHUB_TOKEN-pushed tag would not).
+
+name: Update analyzers & publish
+
+on:
+  workflow_dispatch: {}
+  repository_dispatch:
+    types: [package-analyzers-release]
+
+permissions:
+  contents: write
+
+jobs:
+  bump-and-release:
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4
+        with:
+          submodules: recursive
+          fetch-depth: 0
+          token: ${{ secrets.CLASSIC_PAT }}
+
+      - name: Configure git identity
+        run: |
+          git config user.name  'github-actions[bot]'
+          git config user.email '41898282+github-actions[bot]@users.noreply.github.com'
+
+      - name: Move analyzers to latest upstream
+        id: bump
+        run: |
+          set -euo pipefail
+          before=$(git rev-parse HEAD:NLPPlus/analyzers)
+          git submodule update --remote --recursive NLPPlus/analyzers
+          after=$(git -C NLPPlus/analyzers rev-parse HEAD)
+          echo "before=$before" >> "$GITHUB_OUTPUT"
+          echo "after=$after"   >> "$GITHUB_OUTPUT"
+          if [ "$before" = "$after" ]; then
+            echo "changed=false" >> "$GITHUB_OUTPUT"
+            echo "analyzers already current ($after) — nothing to publish"
+          else
+            echo "changed=true" >> "$GITHUB_OUTPUT"
+            echo "analyzers: $before -> $after"
+          fi
+
+      - name: Compute next version
+        if: steps.bump.outputs.changed == 'true'
+        id: version
+        run: |
+          set -euo pipefail
+          latest=$(git tag --sort=-v:refname \
+                   | grep -E '^v[0-9]+\.[0-9]+\.[0-9]+$' \
+                   | head -n1 || true)
+          latest=${latest:-v0.0.0}
+          IFS='.' read -r major minor patch <<<"${latest#v}"
+          patch=$((patch + 1))
+          new="v${major}.${minor}.${patch}"
+          echo "new=$new" >> "$GITHUB_OUTPUT"
+          echo "Bumping $latest -> $new"
+
+      - name: Commit, tag, push
+        if: steps.bump.outputs.changed == 'true'
+        run: |
+          set -euo pipefail
+          git add NLPPlus/analyzers
+          git commit -m "Bump analyzers submodule to ${{ steps.bump.outputs.after }} (${{ steps.version.outputs.new }})"
+          git tag -a "${{ steps.version.outputs.new }}" -m "Release ${{ steps.version.outputs.new }} (analyzers bump)"
+          git push origin HEAD
+          git push origin "${{ steps.version.outputs.new }}"
+
+      - name: Summary
+        run: |
+          {
+            echo "### Update analyzers & publish"
+            echo "- analyzers changed: \`${{ steps.bump.outputs.changed }}\`"
+            echo "- analyzers: \`${{ steps.bump.outputs.before }}\` → \`${{ steps.bump.outputs.after }}\`"
+            echo "- new tag: \`${{ steps.version.outputs.new }}\`"
+          } >> "$GITHUB_STEP_SUMMARY"

nlpplus-2.0.5/.gitmodules

@@ -0,0 +1,6 @@
+[submodule "nlp-engine"]
+	path = nlp-engine
+	url = https://github.com/VisualText/nlp-engine.git
+[submodule "NLPPlus/analyzers"]
+	path = NLPPlus/analyzers
+	url = https://github.com/VisualText/package-analyzers.git

{nlpplus-2.0.3 → nlpplus-2.0.5}/NLPPlus/__init__.py

@@ -83,16 +83,18 @@ class Engine:
         verbose: bool = False,
         initialize: bool = False,
     ):
+        self._closed = False
         if working_folder is None:
-            # ignore_cleanup_errors=True: the nlp-engine's static `cgerr`
-            # ofstream holds <analyzer>/logs/cgerr.log open for the
-            # lifetime of the process. Python's TemporaryDirectory
-            # cleanup runs during atexit, but the C++ static destructor
-            # that would close cgerr runs *after* atexit (Python returns
-            # control to the OS first). On Windows that ordering means
-            # cleanup hits a still-open file and raises PermissionError.
-            # Swallow it — the OS reclaims the tempdir at process exit
-            # regardless. Tracked engine-side as NLP-ENGINE-523.
+            # ignore_cleanup_errors=True is now defense-in-depth: close()
+            # / __exit__ / __del__ explicitly call self.engine.close()
+            # before TemporaryDirectory.cleanup, which closes the engine's
+            # cgerr.log handle and lets Windows delete the temp dir
+            # cleanly. The flag still catches the corner case of an
+            # interpreter shutdown where __del__ never runs (e.g. crash,
+            # os._exit) — the OS reclaims the tempdir at process exit
+            # regardless, so swallowing the cleanup error keeps stderr
+            # quiet for users who never explicitly close. Engine-side
+            # fix shipped in NLP-ENGINE-523 (engine v3.1.55+).
             self.tmpdir = TemporaryDirectory(
                 prefix="NLPPlus-", ignore_cleanup_errors=True
             )
@@ -118,6 +120,57 @@ class Engine:
             )
         self.engine = NLP_ENGINE(str(self.working_folder), silent=not verbose)
 
+    def close(self):
+        """Tear down the underlying engine and release the working folder.
+
+        Idempotent: safe to call multiple times. After ``close()``, any
+        call to :meth:`analyze`, :meth:`compile`, or :meth:`cloud_compile`
+        is undefined behavior — create a new ``Engine`` instead.
+
+        On Windows in particular, calling ``close()`` (or using ``Engine``
+        as a context manager) is what makes the auto-created
+        ``TemporaryDirectory`` working folder delete cleanly: the engine
+        keeps a file handle on ``<workfolder>/logs/cgerr.log`` open for
+        the lifetime of the C++ instance, and Windows refuses to delete
+        a directory that contains an open file. ``close()`` calls into
+        the C++ engine's ``close()`` (NLP-ENGINE-523, engine v3.1.55+),
+        which releases that handle before the tempdir is removed.
+        """
+        if self._closed:
+            return
+        self._closed = True
+        # Engine.close() is idempotent on the C++ side as of engine
+        # v3.1.55; older engines just no-op the second teardown.
+        try:
+            self.engine.close()
+        except AttributeError:
+            # Pre-3.1.55 binding without the close() method exposed;
+            # fall back to letting __del__ tear it down. The tempdir
+            # cleanup below will still hit the PermissionError on
+            # Windows in that case — same situation as before 2.0.4.
+            pass
+        if self.tmpdir is not None:
+            self.tmpdir.cleanup()
+            self.tmpdir = None
+
+    def __enter__(self):
+        return self
+
+    def __exit__(self, exc_type, exc_value, traceback):
+        self.close()
+        return False
+
+    def __del__(self):
+        # Best-effort: __del__ may run during interpreter shutdown when
+        # modules are being torn down and the bindings module may already
+        # be gone. Swallow anything that goes wrong here; the explicit
+        # close() / context-manager paths are the supported way to get
+        # deterministic cleanup.
+        try:
+            self.close()
+        except Exception:
+            pass
+
     def analyze(self, text: str, analyzer_name: str, develop: bool = False,
                 compiled: bool = False) -> Results:
         """Analyze text with the named analyzer.

{nlpplus-2.0.3 → nlpplus-2.0.5}/NLPPlus/_version.py

@@ -18,7 +18,7 @@ version_tuple: tuple[int | str, ...]
 commit_id: str | None
 __commit_id__: str | None
 
-__version__ = version = '2.0.3'
-__version_tuple__ = version_tuple = (2, 0, 3)
+__version__ = version = '2.0.5'
+__version_tuple__ = version_tuple = (2, 0, 5)
 
-__commit_id__ = commit_id = 'gd30bbfbe8'
+__commit_id__ = commit_id = 'g28161869e'

nlpplus-2.0.5/NLPPlus/analyzers/.github/workflows/dispatch-update-package-analyzers.yml

@@ -0,0 +1,40 @@
+# Install at: VisualText/package-analyzers -> .github/workflows/dispatch-update-package-analyzers.yml
+#
+# Sender in the cross-repo percolation chain (see
+# VisualText/nlp-engine/docs/PERCOLATION.md).
+#
+# When a release tag (v*) is pushed here, ping the repos that embed
+# package-analyzers as a submodule so they refresh their pointer. Both the npm
+# and Python nlpengine packages listen for the `package-analyzers-release` event
+# and open a bump PR (they publish to npm/PyPI with their own versions, so a
+# submodule bump must not auto-publish — a human merges the PR).
+#
+# Requires the shared `CLASSIC_PAT` secret in this repo (the default
+# GITHUB_TOKEN cannot trigger workflows in other repos).
+
+name: Dispatch update-package-analyzers
+
+on:
+  push:
+    tags:
+      - 'v*'
+  workflow_dispatch:
+
+jobs:
+  dispatch:
+    runs-on: ubuntu-latest
+    strategy:
+      fail-fast: false
+      matrix:
+        repo:
+          - VisualText/npm-package-nlpengine   # embeds package-analyzers at path analyzers
+          - VisualText/py-package-nlpengine    # embeds package-analyzers at path NLPPlus/analyzers
+    steps:
+      - name: Trigger ${{ matrix.repo }}
+        uses: peter-evans/repository-dispatch@v2
+        with:
+          token: ${{ secrets.CLASSIC_PAT }}
+          repository: ${{ matrix.repo }}
+          event-type: package-analyzers-release
+          client-payload: '{"tag_name": "${{ github.ref_name }}"}'
+        continue-on-error: true

nlpplus-2.0.5/NLPPlus/analyzers/.github/workflows/tag-on-push.yml

@@ -0,0 +1,86 @@
+# Install at: VisualText/package-analyzers -> .github/workflows/tag-on-push.yml
+#
+# Source of the percolation chain for DIRECT analyzer edits (see
+# VisualText/nlp-engine/docs/PERCOLATION.md).
+#
+# When you edit an analyzer (address-parser, emailaddress, links, telephone) and
+# push to main, this auto-computes the next semver tag and pushes it. Pushing the
+# v* tag fires dispatch-update-package-analyzers.yml, which pings the npm and
+# Python packages to bump + auto-publish.
+#
+# parse-en-us is intentionally NOT in the path filter: parse-en-us changes arrive
+# via update-parse-en-us.yml (driven by parse-en-us-release), which tags itself.
+# Keeping the two paths disjoint avoids double-tagging.
+#
+# Requires the shared CLASSIC_PAT secret (a GITHUB_TOKEN-pushed tag cannot
+# trigger dispatch-update-package-analyzers.yml in turn).
+
+name: Tag on push
+
+on:
+  push:
+    branches: [main]
+    paths:
+      - 'address-parser/**'
+      - 'emailaddress/**'
+      - 'links/**'
+      - 'telephone/**'
+  workflow_dispatch:
+    inputs:
+      bump:
+        description: "Semver component to bump"
+        required: true
+        default: "patch"
+        type: choice
+        options: [patch, minor, major]
+
+permissions:
+  contents: write
+
+jobs:
+  tag:
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4
+        with:
+          fetch-depth: 0
+          token: ${{ secrets.CLASSIC_PAT }}
+
+      - name: Configure git identity
+        run: |
+          git config user.name  'github-actions[bot]'
+          git config user.email '41898282+github-actions[bot]@users.noreply.github.com'
+
+      - name: Compute next version
+        id: version
+        run: |
+          set -euo pipefail
+          bump="${{ inputs.bump }}"
+          bump="${bump:-patch}"   # push events have no inputs
+          latest=$(git tag --sort=-v:refname \
+                   | grep -E '^v[0-9]+\.[0-9]+\.[0-9]+$' \
+                   | head -n1 || true)
+          latest=${latest:-v0.0.0}
+          IFS='.' read -r major minor patch <<<"${latest#v}"
+          case "$bump" in
+            major) major=$((major + 1)); minor=0; patch=0 ;;
+            minor) minor=$((minor + 1)); patch=0 ;;
+            patch) patch=$((patch + 1)) ;;
+          esac
+          new="v${major}.${minor}.${patch}"
+          echo "new=$new" >> "$GITHUB_OUTPUT"
+          echo "Tagging $latest -> $new"
+
+      - name: Tag and push
+        run: |
+          set -euo pipefail
+          git tag -a "${{ steps.version.outputs.new }}" -m "Release ${{ steps.version.outputs.new }}"
+          git push origin "${{ steps.version.outputs.new }}"
+
+      - name: Create GitHub release
+        uses: softprops/action-gh-release@v2
+        with:
+          tag_name: ${{ steps.version.outputs.new }}
+          name: ${{ steps.version.outputs.new }}
+          generate_release_notes: true
+          token: ${{ secrets.CLASSIC_PAT || secrets.GITHUB_TOKEN }}

nlpplus-2.0.5/NLPPlus/analyzers/.github/workflows/update-parse-en-us.yml

@@ -0,0 +1,130 @@
+# Install at: VisualText/package-analyzers -> .github/workflows/update-parse-en-us.yml
+#
+# Listener in the cross-repo percolation chain (see
+# VisualText/nlp-engine/docs/PERCOLATION.md).
+#
+# 1. Pulls the latest commit of the parse-en-us submodule's default branch (main).
+# 2. Computes a new semver tag by bumping the latest vX.Y.Z tag.
+# 3. Commits the submodule pointer change, tags, and pushes.
+# 4. Cuts a GitHub release at the new tag. Pushing the v* tag fires
+#    dispatch-update-package-analyzers.yml, which pings the npm and Python
+#    packages to refresh their analyzers submodule.
+#
+# Triggered by parse-en-us (event-type parse-en-us-release) on its release, or
+# by hand. A repository_dispatch ping carries no inputs, so the bump defaults to
+# patch.
+
+name: Update parse-en-us & Bump Version
+
+on:
+  workflow_dispatch:
+    inputs:
+      bump:
+        description: "Semver component to bump"
+        required: true
+        default: "patch"
+        type: choice
+        options:
+          - patch
+          - minor
+          - major
+      create_release:
+        description: "Create a GitHub release at the new tag"
+        required: true
+        default: true
+        type: boolean
+  repository_dispatch:
+    types: [parse-en-us-release]
+
+permissions:
+  contents: write
+
+jobs:
+  update-and-bump:
+    runs-on: ubuntu-latest
+    steps:
+      - name: Checkout (with submodules and full history)
+        uses: actions/checkout@v4
+        with:
+          fetch-depth: 0
+          submodules: recursive
+          token: ${{ secrets.CLASSIC_PAT || secrets.GITHUB_TOKEN }}
+
+      - name: Configure git identity
+        run: |
+          git config user.name  'github-actions[bot]'
+          git config user.email '41898282+github-actions[bot]@users.noreply.github.com'
+
+      - name: Update parse-en-us to latest main
+        id: submodule
+        run: |
+          set -euo pipefail
+          before=$(git -C parse-en-us rev-parse HEAD)
+          git -C parse-en-us fetch origin main
+          git -C parse-en-us checkout origin/main
+          git -C parse-en-us submodule update --init --recursive
+          after=$(git -C parse-en-us rev-parse HEAD)
+          echo "before=$before" >> "$GITHUB_OUTPUT"
+          echo "after=$after"   >> "$GITHUB_OUTPUT"
+          if [ "$before" = "$after" ]; then
+            echo "changed=false" >> "$GITHUB_OUTPUT"
+            echo "parse-en-us already at $after"
+          else
+            echo "changed=true" >> "$GITHUB_OUTPUT"
+            echo "parse-en-us: $before -> $after"
+          fi
+
+      - name: Compute next version
+        id: version
+        run: |
+          set -euo pipefail
+          bump="${{ inputs.bump }}"
+          bump="${bump:-patch}"   # repository_dispatch has no inputs
+          latest=$(git tag --sort=-v:refname \
+                   | grep -E '^v[0-9]+\.[0-9]+\.[0-9]+$' \
+                   | head -n1 || true)
+          latest=${latest:-v0.0.0}
+          IFS='.' read -r major minor patch <<<"${latest#v}"
+          case "$bump" in
+            major) major=$((major + 1)); minor=0; patch=0 ;;
+            minor) minor=$((minor + 1)); patch=0 ;;
+            patch) patch=$((patch + 1)) ;;
+          esac
+          new="v${major}.${minor}.${patch}"
+          echo "previous=$latest" >> "$GITHUB_OUTPUT"
+          echo "new=$new"         >> "$GITHUB_OUTPUT"
+          echo "Bumping $latest -> $new"
+
+      - name: Commit submodule update
+        if: steps.submodule.outputs.changed == 'true'
+        run: |
+          git add parse-en-us
+          git commit -m "Update parse-en-us to ${{ steps.submodule.outputs.after }} (${{ steps.version.outputs.new }})"
+
+      - name: Tag and push
+        run: |
+          set -euo pipefail
+          git tag -a "${{ steps.version.outputs.new }}" \
+            -m "Release ${{ steps.version.outputs.new }}"
+          git push origin HEAD
+          git push origin "${{ steps.version.outputs.new }}"
+
+      - name: Create GitHub release
+        if: ${{ github.event_name != 'workflow_dispatch' || inputs.create_release }}
+        uses: softprops/action-gh-release@v2
+        with:
+          tag_name: ${{ steps.version.outputs.new }}
+          name: ${{ steps.version.outputs.new }}
+          generate_release_notes: true
+          token: ${{ secrets.CLASSIC_PAT || secrets.GITHUB_TOKEN }}
+
+      - name: Summary
+        run: |
+          {
+            echo "### Update parse-en-us & bump version"
+            echo ""
+            echo "- previous tag: \`${{ steps.version.outputs.previous }}\`"
+            echo "- new tag: \`${{ steps.version.outputs.new }}\`"
+            echo "- parse-en-us changed: \`${{ steps.submodule.outputs.changed }}\`"
+            echo "- parse-en-us: \`${{ steps.submodule.outputs.before }}\` → \`${{ steps.submodule.outputs.after }}\`"
+          } >> "$GITHUB_STEP_SUMMARY"

nlpplus-2.0.5/NLPPlus/analyzers/.gitmodules

@@ -0,0 +1,3 @@
+[submodule "parse-en-us"]
+	path = parse-en-us
+	url = https://github.com/VisualText/parse-en-us.git

nlpplus-2.0.5/NLPPlus/analyzers/parse-en-us/.github/workflows/dispatch-update-parse-en-us.yml

@@ -0,0 +1,37 @@
+# Install at: VisualText/parse-en-us -> .github/workflows/dispatch-update-parse-en-us.yml
+#
+# Release-driven percolation, matching the existing org convention
+# (cf. analyzer-templates/dispatch-update-analyzer-templates.yml): when a
+# release tag is pushed here, ping the repos that embed parse-en-us so they
+# refresh their submodule pointer.
+#
+# Requires the shared `CLASSIC_PAT` secret in this repo (the default
+# GITHUB_TOKEN cannot trigger workflows in other repos).
+
+name: Dispatch update-parse-en-us
+
+on:
+  push:
+    tags:
+      - 'v*'
+  workflow_dispatch:
+
+jobs:
+  dispatch:
+    runs-on: ubuntu-latest
+    strategy:
+      fail-fast: false
+      matrix:
+        repo:
+          - analyzers           # embeds parse-en-us at path parse-en-us
+          - analyzer-templates  # embeds parse-en-us at path parse-en-us
+          - visualtext-files    # embeds parse-en-us at path analyzers/parse-en-us
+    steps:
+      - name: Trigger ${{ matrix.repo }}
+        uses: peter-evans/repository-dispatch@v2
+        with:
+          token: ${{ secrets.CLASSIC_PAT }}
+          repository: VisualText/${{ matrix.repo }}
+          event-type: parse-en-us-release
+          client-payload: '{"tag_name": "${{ github.ref_name }}"}'
+        continue-on-error: true

nlpplus-2.0.5/NLPPlus/analyzers/parse-en-us/README.md

@@ -0,0 +1,15 @@
+# Full USA English Parser
+
+This is a full English Parser for USA English created by Amnon Meyers
+
+It was developed over the period of many years in a bottom up manner. It uses the older Wordnet dictionary for parts of speech and uses only syntax, no semantics.
+
+## Starting Template
+
+It is recommended that if full parsing of English is needed in an analyzer, to start with this. It has been used as the basis for numerous text analyzers throughout the years.
+
+## Cross-repo release automation
+
+This repo participates in the VisualText cross-repo release "percolation"
+system: submodule bumps flow downstream automatically via `repository_dispatch`.
+See **[nlp-engine/docs/PERCOLATION.md](https://github.com/VisualText/nlp-engine/blob/master/docs/PERCOLATION.md)** for the full map.

nlpplus-2.0.5/NLPPlus/analyzers/parse-en-us/input/business.txt

@@ -0,0 +1,37 @@
+LOS ANGELES (NWI) - Thetext company Text
+Analysis International Inc. (TAI) has bought the American
+Medical Records Processing (AMRP) for more than $130
+million dollars. Avi Meyers, CEO and president of TAI,
+said on Friday that they bought the Los Angeles-based AMRP
+for $120 million in cash and $14.3 million in assumed debt.
+
+Founded in 1972, AMRP is the largest text processing firm
+in the country for medical records and journals and has
+been using TAI's VisualText technology almost exclusively
+for 3 years.
+
+The sale of the medical records processing company for more
+than $134.3 million is above analysts' predictions but is
+not surprising given TAI's strong position in the market
+place.
+
+In 2006 alone, TAI has acquired MD Files, New Hampshire
+Medical Inc., and Carezani Med Company of Maryland for a
+total of 160 million dollars.
+
+With this last purchase, TAI has substantially strengthened
+its hold on the medical text processing field and hopes to
+centralize the processing of medical records nationally.
+
+With its latest purchase of the medical records processing
+company, TAI's stock is up 4% from $58.33 a share to
+$60.66.
+
+TAI's stock has recently remained steady due to its lawsuit
+with MiasmaCraft whose text processing IDE Miasma-Text bears
+a striking resemblance to TAI's VisualText.
+
+TAI is vigorously defending its patent on VisualText,
+NLP++, and the Conceptual Grammar.
+
+More than they know.

nlpplus-2.0.5/NLPPlus/analyzers/parse-en-us/input/doj.txt

@@ -0,0 +1,7 @@
+Also unsealed today in federal court in the Eastern District of New York was the guilty plea of
+Tim Leissner, 48, the former Southeast Asia Chairman and participating managing director of the
+Financial Institution, to a two-count criminal information charging Leissner with conspiring to
+launder money and conspiring to violate the FCPA by both paying bribes to various Malaysian and
+Abu Dhabi officials and circumventing the internal accounting controls of the Financial
+Institution while he was employed by it. According to court filings, Leissner has been ordered
+to forfeit $43.7 million as a result of his crimes.

nlpplus-2.0.5/NLPPlus/analyzers/parse-en-us/input/test.txt

@@ -0,0 +1 @@
+The quick brown fox jumped over the lazy boy.

nlpplus-2.0.3/NLPPlus/analyzers/parse-en-us/kb/user/en-full.dict → nlpplus-2.0.5/NLPPlus/analyzers/parse-en-us/kb/user/all.dict

@@ -1,4 +1,3 @@
-# Full English Dictionary
 aah pos=verb
 aah pos=int
 aahed pos=verb

{nlpplus-2.0.3 → nlpplus-2.0.5}/NLPPlus/analyzers/parse-en-us/spec/semfuns.nlp

@@ -302,7 +302,11 @@ if (L("nsenses"))	# 02/08/11 AM.
 if (L("matchcons") = pnvar(L("v"),"MATCHCONS"))		# 03/17/14 AM.
   {
   pnreplaceval(L("vg"),"MATCHCONS",L("matchcons"));	# 03/17/14 AM.
-  PRETTYCONShier(L("matchcons"),G("mhdump"));
+  # PRETTYCONShier(L("matchcons"),G("mhdump"));	# Undefined function.
+  # The interpreter tolerated this (the if-condition is rarely true,
+  # so the call rarely fired), but the C++ compile path can't resolve
+  # the symbol at link time and the build fails. Commented out pending
+  # a decision on whether to define the function or drop the block.
   }
 }
 

{nlpplus-2.0.3 → nlpplus-2.0.5}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.2
 Name: NLPPlus
-Version: 2.0.3
+Version: 2.0.5
 Summary: Python extension for the NLP++ text analysis engine
 Keywords: nlp,text
 Author-Email: David De Hilster <contact@visualtext.org>

{nlpplus-2.0.3 → nlpplus-2.0.5}/bindings.cpp

@@ -119,5 +119,20 @@ NB_MODULE(bindings, m) {
              "Generates <analyzer>/run/*.cpp and <analyzer>/kb/*.cpp\n"
              "(or just <analyzer>/kb/*.cpp if `kbOnly=True`).  Those\n"
              "still need to be built into shared libraries before\n"
-             "`analyze(..., compiled=True)` can load them.");
+             "`analyze(..., compiled=True)` can load them.")
+        // NLP_ENGINE has two close() overloads: close() and
+        // close(_TCHAR *analyzer). Cast to pick the nullary one
+        // (we want the global teardown, not the per-analyzer one).
+        .def("close",
+             static_cast<int (NLP_ENGINE::*)()>(&NLP_ENGINE::close),
+             "Tear down the engine's VTRun runtime and release the open\n"
+             "<workfolder>/logs/cgerr.log handle. Safe to call multiple\n"
+             "times (engine v3.1.55+ NLP-ENGINE-523 made close()\n"
+             "idempotent). After close() returns, any subsequent\n"
+             "analyze()/compile() call on this instance is undefined\n"
+             "behavior. NLPPlus.Engine.close() / __exit__ / __del__\n"
+             "calls this before deleting the TemporaryDirectory backing\n"
+             "the working folder; this is what makes the temp-dir\n"
+             "cleanup safe on Windows where an open file handle would\n"
+             "otherwise block the rmtree.");
 }