Annotation extractor function (fix #26)

code-hike · Oct 11, 2023 · 567aadb · 567aadb · vercel · Oct 11, 2023
1 parent 677d2d8
commit 567aadb
Show file tree

Hide file tree

Showing 12 changed files with 483 additions and 15 deletions.
diff --git a/.changeset/clever-cheetahs-fry.md b/.changeset/clever-cheetahs-fry.md
@@ -0,0 +1,5 @@
+---
+"@code-hike/lighter": minor
+---
+
+Add annotation extractor function
diff --git a/lib/dist/browser.esm.mjs b/lib/dist/browser.esm.mjs
diff --git a/lib/dist/index.cjs.js b/lib/dist/index.cjs.js
diff --git a/lib/dist/index.d.ts b/lib/dist/index.d.ts
@@ -48,6 +48,12 @@ type Annotation = {
     query?: string;
     ranges: CodeRange[];
 };
+type AnnotationData = {
+    name: string;
+    rangeString: string;
+    query?: string;
+};
+type AnnotationExtractor = string[] | ((comment: string) => null | AnnotationData);
 
 type Token = {
     content: string;
@@ -113,7 +119,7 @@ declare function highlight(code: string, lang: LanguageAlias, themeOrThemeName?:
 declare function highlight(code: string, lang: LanguageAlias, themeOrThemeName: Theme, config: AnnotatedConfig): Promise<AnnotatedLighterResult>;
 declare function highlightSync(code: string, lang: LanguageAlias, themeOrThemeName?: Theme, config?: Config): LighterResult;
 declare function highlightSync(code: string, lang: LanguageAlias, themeOrThemeName: Theme, config: AnnotatedConfig): AnnotatedLighterResult;
-declare function extractAnnotations(code: string, lang: LanguageAlias, annotationNames?: string[]): Promise<{
+declare function extractAnnotations(code: string, lang: LanguageAlias, annotationExtractor?: AnnotationExtractor): Promise<{
     code: string;
     annotations: Annotation[];
 }>;

diff --git a/lib/dist/index.esm.mjs b/lib/dist/index.esm.mjs
diff --git a/lib/dist/worker.esm.mjs b/lib/dist/worker.esm.mjs
diff --git a/lib/src/comments.ts b/lib/src/comments.ts
@@ -32,11 +32,21 @@ export type Annotation = {
   ranges: CodeRange[];
 };
 
+export type AnnotationData = {
+  name: string;
+  rangeString: string;
+  query?: string;
+};
+
+export type AnnotationExtractor =
+  | string[]
+  | ((comment: string) => null | AnnotationData);
+
 export function extractCommentsFromCode(
   code: string,
   grammar: IGrammar,
   lang: string,
-  annotationNames: string[]
+  annotationExtractor: AnnotationExtractor
 ) {
   const lines = !grammar
     ? highlightText(code)
@@ -49,7 +59,7 @@ export function extractCommentsFromCode(
     .map((line) => {
       const { annotations, lineWithoutComments } = getAnnotationsFromLine(
         line,
-        annotationNames,
+        annotationExtractor,
         lineNumber
       );
 
@@ -81,7 +91,7 @@ export function extractCommentsFromCode(
 
 function getAnnotationsFromLine(
   tokens: Token[],
-  names: string[],
+  annotationExtractor: AnnotationExtractor,
   lineNumber: number
 ) {
   // if no punctuation return empty
@@ -106,12 +116,17 @@ function getAnnotationsFromLine(
       continue;
     }
 
-    const { name, query, rangeString } = getAnnotationData(token.content);
-    if (!names.includes(name)) {
+    const annotationData =
+      typeof annotationExtractor === "function"
+        ? annotationExtractor(token.content)
+        : getAnnotationDataFromNames(token.content, annotationExtractor);
+
+    if (!annotationData) {
       // a comment, but not an annotation
       i++;
       continue;
     }
+    const { name, query, rangeString } = annotationData;
 
     // we have an annotation
     const prevToken = tokens[i - 1];
@@ -156,11 +171,16 @@ function getAnnotationsFromLine(
   };
 }
 
-function getAnnotationData(content: string) {
+function getAnnotationDataFromNames(content: string, names: string[]) {
   const regex = /\s*([\w-]+)?(\([^\)]*\)|\[[^\]]*\])?(.*)$/;
   const match = content.match(regex);
   const name = match[1];
   const rangeString = match[2];
   const query = match[3]?.trim();
+
+  if (!names.includes(name)) {
+    return null;
+  }
+
   return { name, rangeString, query };
 }
diff --git a/lib/src/index.ts b/lib/src/index.ts
@@ -17,7 +17,11 @@ import {
   preloadGrammars,
   highlightText,
 } from "./highlighter";
-import { Annotation, extractCommentsFromCode } from "./comments";
+import {
+  Annotation,
+  AnnotationExtractor,
+  extractCommentsFromCode,
+} from "./comments";
 import {
   applyAnnotations,
   Lines,
@@ -174,9 +178,9 @@ export function highlightSync(
 export async function extractAnnotations(
   code: string,
   lang: LanguageAlias,
-  annotationNames: string[] = []
+  annotationExtractor?: AnnotationExtractor
 ) {
-  if (annotationNames.length === 0) {
+  if (!annotationExtractor) {
     return { code, annotations: [] };
   }
 
@@ -187,7 +191,7 @@ export async function extractAnnotations(
     code,
     grammar,
     lang,
-    annotationNames
+    annotationExtractor
   );
 
   return { code: newCode, annotations };

diff --git a/lib/test/__snapshots__/browser.test.ts.snap b/lib/test/__snapshots__/browser.test.ts.snap
@@ -430,6 +430,140 @@ exports[`extract annottations from txt 2`] = `
 }
 `;
 
+exports[`extract annottations with prefix 1`] = `
+{
+  "annotations": [
+    {
+      "name": "xy",
+      "query": "bar",
+      "ranges": [
+        {
+          "fromColumn": 3,
+          "lineNumber": 3,
+          "toColumn": 5,
+        },
+      ],
+    },
+  ],
+  "code": "// xyz[3:5] foo
+const x = 1;
+const y = 2;",
+}
+`;
+
+exports[`extract annottations with prefix 2`] = `
+{
+  "lang": "javascript",
+  "lines": [
+    {
+      "lineNumber": 1,
+      "tokens": [
+        {
+          "content": "// xyz[3:5] foo",
+          "style": {
+            "color": "#6A9955",
+          },
+        },
+      ],
+    },
+    {
+      "lineNumber": 2,
+      "tokens": [
+        {
+          "content": "const ",
+          "style": {
+            "color": "#569CD6",
+          },
+        },
+        {
+          "content": "x",
+          "style": {
+            "color": "#4FC1FF",
+          },
+        },
+        {
+          "content": " = ",
+          "style": {
+            "color": "#D4D4D4",
+          },
+        },
+        {
+          "content": "1",
+          "style": {
+            "color": "#B5CEA8",
+          },
+        },
+        {
+          "content": ";",
+          "style": {
+            "color": "#D4D4D4",
+          },
+        },
+      ],
+    },
+    {
+      "lineNumber": 3,
+      "tokens": [
+        {
+          "content": "co",
+          "style": {
+            "color": "#569CD6",
+          },
+        },
+        {
+          "annotationName": "xy",
+          "annotationQuery": "bar",
+          "fromColumn": 3,
+          "toColumn": 5,
+          "tokens": [
+            {
+              "content": "nst",
+              "style": {
+                "color": "#569CD6",
+              },
+            },
+          ],
+        },
+        {
+          "content": " ",
+          "style": {
+            "color": "#569CD6",
+          },
+        },
+        {
+          "content": "y",
+          "style": {
+            "color": "#4FC1FF",
+          },
+        },
+        {
+          "content": " = ",
+          "style": {
+            "color": "#D4D4D4",
+          },
+        },
+        {
+          "content": "2",
+          "style": {
+            "color": "#B5CEA8",
+          },
+        },
+        {
+          "content": ";",
+          "style": {
+            "color": "#D4D4D4",
+          },
+        },
+      ],
+    },
+  ],
+  "style": {
+    "background": "#1E1E1E",
+    "color": "#D4D4D4",
+  },
+}
+`;
+
 exports[`highlight html with theme 1`] = `
 {
   "lang": "html",