Merge pull request #64 from rasendubi/feat-sub-superscripts

feat(uniorg-parse): support useSubSuperscripts options
rasendubi · Jan 29, 2023 · 5d8999d · 5d8999d
2 parents 6617c95 + 2325611
commit 5d8999d
Show file tree

Hide file tree

Showing 5 changed files with 322 additions and 28 deletions.
diff --git a/packages/uniorg-parse/src/__snapshots__/parser.spec.ts.snap b/packages/uniorg-parse/src/__snapshots__/parser.spec.ts.snap
@@ -3239,6 +3239,210 @@ children:
         value: "hello\\n#+BLAH[hi]: heh\\nhi\\n"
 `;
 
+exports[`org/parser options.useSubSuperscript = {} ignores subscript with number 1`] = `
+type: "org-data"
+contentsBegin: 0
+contentsEnd: 3
+children:
+  - type: "paragraph"
+    affiliated: {}
+    contentsBegin: 0
+    contentsEnd: 3
+    children:
+      - type: "text"
+        value: "x_2"
+`;
+
+exports[`org/parser options.useSubSuperscript = {} ignores superscript with number 1`] = `
+type: "org-data"
+contentsBegin: 0
+contentsEnd: 3
+children:
+  - type: "paragraph"
+    affiliated: {}
+    contentsBegin: 0
+    contentsEnd: 3
+    children:
+      - type: "text"
+        value: "x^2"
+`;
+
+exports[`org/parser options.useSubSuperscript = {} parses subscript with braces 1`] = `
+type: "org-data"
+contentsBegin: 0
+contentsEnd: 5
+children:
+  - type: "paragraph"
+    affiliated: {}
+    contentsBegin: 0
+    contentsEnd: 5
+    children:
+      - type: "text"
+        value: "x"
+      - type: "subscript"
+        contentsBegin: 3
+        contentsEnd: 4
+        children:
+          - type: "text"
+            value: "2"
+`;
+
+exports[`org/parser options.useSubSuperscript = {} parses superscript with braces 1`] = `
+type: "org-data"
+contentsBegin: 0
+contentsEnd: 5
+children:
+  - type: "paragraph"
+    affiliated: {}
+    contentsBegin: 0
+    contentsEnd: 5
+    children:
+      - type: "text"
+        value: "x"
+      - type: "superscript"
+        contentsBegin: 3
+        contentsEnd: 4
+        children:
+          - type: "text"
+            value: "2"
+`;
+
+exports[`org/parser options.useSubSuperscript = false ignores subscript with braces 1`] = `
+type: "org-data"
+contentsBegin: 0
+contentsEnd: 5
+children:
+  - type: "paragraph"
+    affiliated: {}
+    contentsBegin: 0
+    contentsEnd: 5
+    children:
+      - type: "text"
+        value: "x_{2}"
+`;
+
+exports[`org/parser options.useSubSuperscript = false ignores subscript with number 1`] = `
+type: "org-data"
+contentsBegin: 0
+contentsEnd: 3
+children:
+  - type: "paragraph"
+    affiliated: {}
+    contentsBegin: 0
+    contentsEnd: 3
+    children:
+      - type: "text"
+        value: "x_2"
+`;
+
+exports[`org/parser options.useSubSuperscript = false ignores superscript with braces 1`] = `
+type: "org-data"
+contentsBegin: 0
+contentsEnd: 5
+children:
+  - type: "paragraph"
+    affiliated: {}
+    contentsBegin: 0
+    contentsEnd: 5
+    children:
+      - type: "text"
+        value: "x^{2}"
+`;
+
+exports[`org/parser options.useSubSuperscript = false ignores superscript with number 1`] = `
+type: "org-data"
+contentsBegin: 0
+contentsEnd: 3
+children:
+  - type: "paragraph"
+    affiliated: {}
+    contentsBegin: 0
+    contentsEnd: 3
+    children:
+      - type: "text"
+        value: "x^2"
+`;
+
+exports[`org/parser options.useSubSuperscript = true parses subscript with braces 1`] = `
+type: "org-data"
+contentsBegin: 0
+contentsEnd: 5
+children:
+  - type: "paragraph"
+    affiliated: {}
+    contentsBegin: 0
+    contentsEnd: 5
+    children:
+      - type: "text"
+        value: "x"
+      - type: "subscript"
+        contentsBegin: 3
+        contentsEnd: 4
+        children:
+          - type: "text"
+            value: "2"
+`;
+
+exports[`org/parser options.useSubSuperscript = true parses subscript with number 1`] = `
+type: "org-data"
+contentsBegin: 0
+contentsEnd: 3
+children:
+  - type: "paragraph"
+    affiliated: {}
+    contentsBegin: 0
+    contentsEnd: 3
+    children:
+      - type: "text"
+        value: "x"
+      - type: "subscript"
+        contentsBegin: 2
+        contentsEnd: 3
+        children:
+          - type: "text"
+            value: "2"
+`;
+
+exports[`org/parser options.useSubSuperscript = true parses superscript with braces 1`] = `
+type: "org-data"
+contentsBegin: 0
+contentsEnd: 5
+children:
+  - type: "paragraph"
+    affiliated: {}
+    contentsBegin: 0
+    contentsEnd: 5
+    children:
+      - type: "text"
+        value: "x"
+      - type: "superscript"
+        contentsBegin: 3
+        contentsEnd: 4
+        children:
+          - type: "text"
+            value: "2"
+`;
+
+exports[`org/parser options.useSubSuperscript = true parses superscript with number 1`] = `
+type: "org-data"
+contentsBegin: 0
+contentsEnd: 3
+children:
+  - type: "paragraph"
+    affiliated: {}
+    contentsBegin: 0
+    contentsEnd: 3
+    children:
+      - type: "text"
+        value: "x"
+      - type: "superscript"
+        contentsBegin: 2
+        contentsEnd: 3
+        children:
+          - type: "text"
+            value: "2"
+`;
+
 exports[`org/parser paragraph split by empty line 1`] = `
 type: "org-data"
 contentsBegin: 0

diff --git a/packages/uniorg-parse/src/parse-options.ts b/packages/uniorg-parse/src/parse-options.ts
@@ -4,6 +4,14 @@ export interface ParseOptions {
    * plain list of keywords.
    */
   todoKeywords: string[];
+  /**
+   * Same as `org-use-sub-superscripts` in Emacs.
+   *
+   * - `true` (default) - parse sub-/superscripts
+   * - `false` - do not parse sub-/superscripts
+   * - `'{}'` - only parse sub-/superscripts when enclosed in braces
+   */
+  useSubSuperscripts: true | false | '{}';
   /**
    * Allows overriding parameters for emphasis regex. Corresponds to
    * `org-emphasis-regex-components` in Emacs.
@@ -29,6 +37,7 @@ export interface ParseOptions {
 
 export const defaultOptions: ParseOptions = {
   todoKeywords: ['TODO', 'DONE'],
+  useSubSuperscripts: true,
   emphasisRegexpComponents: {
     // deviates from org mode default to allow ndash, mdash, and
     // quotes (’“”)

diff --git a/packages/uniorg-parse/src/parser.spec.ts b/packages/uniorg-parse/src/parser.spec.ts
@@ -671,6 +671,53 @@ not a block
     });
   });
 
+  describe('options.useSubSuperscript', () => {
+    describe('= true', () => {
+      itParses('parses superscript with number', 'x^2', {
+        useSubSuperscripts: true,
+      });
+      itParses('parses subscript with number', 'x_2', {
+        useSubSuperscripts: true,
+      });
+      itParses('parses superscript with braces', 'x^{2}', {
+        useSubSuperscripts: true,
+      });
+      itParses('parses subscript with braces', 'x_{2}', {
+        useSubSuperscripts: true,
+      });
+    });
+
+    describe('= false', () => {
+      itParses('ignores superscript with number', 'x^2', {
+        useSubSuperscripts: false,
+      });
+      itParses('ignores subscript with number', 'x_2', {
+        useSubSuperscripts: false,
+      });
+      itParses('ignores superscript with braces', 'x^{2}', {
+        useSubSuperscripts: false,
+      });
+      itParses('ignores subscript with braces', 'x_{2}', {
+        useSubSuperscripts: false,
+      });
+    });
+
+    describe('= {}', () => {
+      itParses('ignores superscript with number', 'x^2', {
+        useSubSuperscripts: '{}',
+      });
+      itParses('ignores subscript with number', 'x_2', {
+        useSubSuperscripts: '{}',
+      });
+      itParses('parses superscript with braces', 'x^{2}', {
+        useSubSuperscripts: '{}',
+      });
+      itParses('parses subscript with braces', 'x_{2}', {
+        useSubSuperscripts: '{}',
+      });
+    });
+  });
+
   itParses(
     'table',
     `

diff --git a/packages/uniorg-parse/src/parser.ts b/packages/uniorg-parse/src/parser.ts
@@ -1393,9 +1393,19 @@ class Parser {
   // Object parsers.
 
   private parseSuperscript(): Superscript | null {
+    if (!this.options.useSubSuperscripts) {
+      return null;
+    }
+
     this.r.backoff(1); // backoff by one, to match previous char (should be non-space)
     const start = this.r.offset();
-    const m = this.r.advance(this.r.lookingAt(this.re.subsuperscriptRe()));
+    const m = this.r.advance(
+      this.r.lookingAt(
+        this.options.useSubSuperscripts === '{}'
+          ? this.re.matchSubstringWithBracesRegex()
+          : this.re.matchSubstringRegex()
+      )
+    );
     if (!m) return null;
 
     const inside = m.groups!['inBraces'] || m.groups!['inBrackets'];
@@ -1409,9 +1419,19 @@ class Parser {
   }
 
   private parseSubscript(): Subscript | null {
+    if (!this.options.useSubSuperscripts) {
+      return null;
+    }
+
     this.r.backoff(1); // backoff by one, to match previous char (should be non-space)
     const start = this.r.offset();
-    const m = this.r.advance(this.r.lookingAt(this.re.subsuperscriptRe()));
+    const m = this.r.advance(
+      this.r.lookingAt(
+        this.options.useSubSuperscripts === '{}'
+          ? this.re.matchSubstringWithBracesRegex()
+          : this.re.matchSubstringRegex()
+      )
+    );
     if (!m) return null;
 
     const inside = m.groups!['inBraces'] || m.groups!['inBrackets'];