fix!: Update how words are split (#2033)

streetsidesoftware · Mar 14, 2023 · cae637e · cae637e
1 parent 14bc675
commit cae637e
Show file tree

Hide file tree

Showing 20 changed files with 74,054 additions and 1,233 deletions.
diff --git a/cSpell.json b/cSpell.json
@@ -67,6 +67,7 @@
         "checksum.txt",
         "cSpell.json",
         "cspell.json",
+        "dictionaries/*/dict/**/*.txt",
         "dictionaries/*/src/**",
         "generator-cspell-dicts/generators/app/templates/**",
         "lerna.json",

diff --git a/dictionaries/cpp/cspell-ext.json b/dictionaries/cpp/cspell-ext.json
@@ -6,14 +6,19 @@
     "readonly": true,
     // List of dictionary files to add to the global list of dictionaries
     "dictionaryDefinitions": [
+        {
+            "name": "cpp-legacy",
+            "path": "./dict/cpp-legacy.txt",
+            "description": "Legacy C/C++ Keywords and common library functions."
+        },
         {
             "name": "cpp",
-            "path": "./cpp.txt.gz",
+            "path": "./dict/cpp.txt",
             "description": "C/C++ Keywords and common library functions."
         },
         {
             "name": "cpp-refined",
-            "path": "./cpp-refined.txt.gz",
+            "path": "./dict/cpp-refined.txt",
             "description": "Refined list of C/C++ Keywords and common library functions."
         }
     ],

diff --git a/dictionaries/cpp/cspell-tools.config.yaml b/dictionaries/cpp/cspell-tools.config.yaml
@@ -0,0 +1,50 @@
+# yaml-language-server: $schema=https://raw.githubusercontent.com/streetsidesoftware/cspell/main/packages/cspell-tools/cspell-tools.config.schema.json
+
+targets:
+  - name: "cpp"
+    sources:
+      - filename: src/cpp.txt
+        split: legacy
+        allowedSplitWords:
+          - ../en_US/en_US.trie
+          - ../software-terms/dict/softwareTerms.txt
+          - src/cpp.txt
+          - src/allowed-terms.txt
+          - src/compiler-clang-attributes.txt
+          - src/compiler-gcc.txt
+          - src/compiler-msvc.txt
+          - src/ecosystem.txt
+          - src/lang-jargon.txt
+          - src/lang-keywords.txt
+          - src/people.txt
+          - src/stdlib-c.txt
+          - src/stdlib-cerrno.txt
+          - src/stdlib-cmath.txt
+          - src/stdlib-cpp.txt
+          - src/template-strings.txt
+      - filename: src/allowed-terms.txt
+    format: plaintext
+    targetDirectory: "./dict"
+    generateNonStrict: false
+    compress: false
+    sort: true
+  - name: "cpp-legacy"
+    sources:
+      - filename: src/cpp.txt
+        split: legacy
+    format: plaintext
+    targetDirectory: "./dict"
+    generateNonStrict: false
+    compress: false
+  - name: cpp-refined
+    sources:
+      - listFile: source-files-cpp-refined.txt
+        split: true
+        # allowedSplitWords:
+        #   - ../en_US/en_US.trie
+        #   - ../software-terms/dict/softwareTerms.txt
+        #   - src/allowed-terms.txt
+    format: plaintext
+    targetDirectory: "./dict"
+    generateNonStrict: false
+    compress: false
diff --git a/dictionaries/cpp/cspell.json b/dictionaries/cpp/cspell.json
@@ -1,6 +1,6 @@
 {
     "files": ["**/*.{md,txt}"],
-    "ignorePaths": ["/src/cpp.txt"],
+    "ignorePaths": ["/src/cpp.txt", "cpp-legacy.txt", "dict"],
     "dictionaries": ["cpp", "cpp-refined"],
     "import": ["./cspell-ext.json"],
     "ignoreWords": [],