@wdprlib/parser 2.0.9 → 2.0.10

package/dist/index.cjs

@@ -646,7 +646,139 @@ var BLOCK_START_TOKENS = [
   "CLEAR_FLOAT_RIGHT"
 ];
 
+// packages/parser/src/parser/rules/utils.ts
+var SAFE_ATTRIBUTES = new Set([
+  "accept",
+  "align",
+  "alt",
+  "autocapitalize",
+  "autoplay",
+  "background",
+  "bgcolor",
+  "border",
+  "buffered",
+  "checked",
+  "cite",
+  "class",
+  "cols",
+  "colspan",
+  "contenteditable",
+  "controls",
+  "coords",
+  "datetime",
+  "decoding",
+  "default",
+  "dir",
+  "dirname",
+  "disabled",
+  "download",
+  "draggable",
+  "for",
+  "form",
+  "headers",
+  "height",
+  "hidden",
+  "high",
+  "href",
+  "hreflang",
+  "id",
+  "inputmode",
+  "ismap",
+  "itemprop",
+  "kind",
+  "label",
+  "lang",
+  "list",
+  "loop",
+  "low",
+  "max",
+  "maxlength",
+  "min",
+  "minlength",
+  "multiple",
+  "muted",
+  "name",
+  "optimum",
+  "pattern",
+  "placeholder",
+  "poster",
+  "preload",
+  "readonly",
+  "required",
+  "reversed",
+  "role",
+  "rows",
+  "rowspan",
+  "scope",
+  "selected",
+  "shape",
+  "size",
+  "sizes",
+  "span",
+  "spellcheck",
+  "src",
+  "srclang",
+  "srcset",
+  "start",
+  "step",
+  "style",
+  "tabindex",
+  "target",
+  "title",
+  "translate",
+  "type",
+  "usemap",
+  "value",
+  "width",
+  "wrap"
+]);
+function filterUnsafeAttributes(attrs) {
+  const result = {};
+  for (const [key, value] of Object.entries(attrs)) {
+    const lower = key.toLowerCase();
+    if (lower.startsWith("on"))
+      continue;
+    if (lower.startsWith("aria-") || lower.startsWith("data-")) {
+      result[key] = value;
+      continue;
+    }
+    if (!SAFE_ATTRIBUTES.has(lower))
+      continue;
+    if (lower === "id") {
+      result[key] = value.startsWith("u-") ? value : `u-${value}`;
+      continue;
+    }
+    result[key] = value;
+  }
+  return result;
+}
+function parseBlockName(ctx, startPos) {
+  let pos = startPos;
+  let consumed = 0;
+  const token = ctx.tokens[pos];
+  if (!token || token.type !== "TEXT" && token.type !== "IDENTIFIER") {
+    return null;
+  }
+  let name = token.value.toLowerCase();
+  consumed++;
+  pos++;
+  if (ctx.tokens[pos]?.type === "UNDERSCORE") {
+    name += "_";
+    consumed++;
+  }
+  return { name, consumed };
+}
+
 // packages/parser/src/parser/rules/inline/utils.ts
+function isExcludedBlockToken(ctx, tokenPos) {
+  if (!ctx.excludedBlockNames?.size)
+    return false;
+  const token = ctx.tokens[tokenPos];
+  if (token?.type !== "BLOCK_OPEN" && token?.type !== "BLOCK_END_OPEN")
+    return false;
+  const nameResult = parseBlockName(ctx, tokenPos + 1);
+  return nameResult !== null && ctx.excludedBlockNames.has(nameResult.name);
+}
 function canApplyInlineRule(rule, token) {
   if (rule.startTokens.length === 0) {
     return true;
@@ -738,7 +870,8 @@ function parseInlineUntil(ctx, endType) {
           isInvalidHeading = true;
         }
       }
-      const isBlockStart = nextMeaningfulToken && BLOCK_START_TOKENS.includes(nextMeaningfulToken.type) && nextMeaningfulToken.lineStart && !isOrphanCloseSpan && !isAnchorName && !isInvalidBlockOpen && !isInvalidHeading;
+      const isExcludedBlock = (nextMeaningfulToken?.type === "BLOCK_OPEN" || nextMeaningfulToken?.type === "BLOCK_END_OPEN") && isExcludedBlockToken(ctx, pos + lookAhead);
+      const isBlockStart = nextMeaningfulToken && BLOCK_START_TOKENS.includes(nextMeaningfulToken.type) && nextMeaningfulToken.lineStart && !isOrphanCloseSpan && !isAnchorName && !isInvalidBlockOpen && !isInvalidHeading && !isExcludedBlock;
       if (!nextMeaningfulToken || nextMeaningfulToken.type === "NEWLINE" || nextMeaningfulToken.type === "EOF" || isBlockStart) {
         if (isBlockStart && nodes.length > 0) {
           const nextPos = pos + lookAhead;
@@ -1076,129 +1209,6 @@ function buildListData(topLtype, list) {
   };
 }
 
-// packages/parser/src/parser/rules/utils.ts
-var SAFE_ATTRIBUTES = new Set([
-  "accept",
-  "align",
-  "alt",
-  "autocapitalize",
-  "autoplay",
-  "background",
-  "bgcolor",
-  "border",
-  "buffered",
-  "checked",
-  "cite",
-  "class",
-  "cols",
-  "colspan",
-  "contenteditable",
-  "controls",
-  "coords",
-  "datetime",
-  "decoding",
-  "default",
-  "dir",
-  "dirname",
-  "disabled",
-  "download",
-  "draggable",
-  "for",
-  "form",
-  "headers",
-  "height",
-  "hidden",
-  "high",
-  "href",
-  "hreflang",
-  "id",
-  "inputmode",
-  "ismap",
-  "itemprop",
-  "kind",
-  "label",
-  "lang",
-  "list",
-  "loop",
-  "low",
-  "max",
-  "maxlength",
-  "min",
-  "minlength",
-  "multiple",
-  "muted",
-  "name",
-  "optimum",
-  "pattern",
-  "placeholder",
-  "poster",
-  "preload",
-  "readonly",
-  "required",
-  "reversed",
-  "role",
-  "rows",
-  "rowspan",
-  "scope",
-  "selected",
-  "shape",
-  "size",
-  "sizes",
-  "span",
-  "spellcheck",
-  "src",
-  "srclang",
-  "srcset",
-  "start",
-  "step",
-  "style",
-  "tabindex",
-  "target",
-  "title",
-  "translate",
-  "type",
-  "usemap",
-  "value",
-  "width",
-  "wrap"
-]);
-function filterUnsafeAttributes(attrs) {
-  const result = {};
-  for (const [key, value] of Object.entries(attrs)) {
-    const lower = key.toLowerCase();
-    if (lower.startsWith("on"))
-      continue;
-    if (lower.startsWith("aria-") || lower.startsWith("data-")) {
-      result[key] = value;
-      continue;
-    }
-    if (!SAFE_ATTRIBUTES.has(lower))
-      continue;
-    if (lower === "id") {
-      result[key] = value.startsWith("u-") ? value : `u-${value}`;
-      continue;
-    }
-    result[key] = value;
-  }
-  return result;
-}
-function parseBlockName(ctx, startPos) {
-  let pos = startPos;
-  let consumed = 0;
-  const token = ctx.tokens[pos];
-  if (!token || token.type !== "TEXT" && token.type !== "IDENTIFIER") {
-    return null;
-  }
-  let name = token.value.toLowerCase();
-  consumed++;
-  pos++;
-  if (ctx.tokens[pos]?.type === "UNDERSCORE") {
-    name += "_";
-    consumed++;
-  }
-  return { name, consumed };
-}
-
 // packages/parser/src/parser/rules/block/utils.ts
 function canApplyBlockRule(rule, token) {
   if (rule.requiresLineStart && !token.lineStart) {
@@ -1209,11 +1219,13 @@ function canApplyBlockRule(rule, token) {
   }
   return rule.startTokens.includes(token.type);
 }
-function parseBlocksUntil(ctx, closeCondition) {
+function parseBlocksUntil(ctx, closeCondition, options) {
   const elements = [];
   let consumed = 0;
   let pos = ctx.pos;
-  const { blockRules, blockFallbackRule } = ctx;
+  const excluded = options?.excludedBlockNames;
+  const blockRules = excluded ? ctx.blockRules.filter((r) => !excluded.has(r.name)) : ctx.blockRules;
+  const { blockFallbackRule } = ctx;
   while (pos < ctx.tokens.length) {
     const token = ctx.tokens[pos];
     if (!token || token.type === "EOF") {
@@ -1234,7 +1246,13 @@ function parseBlocksUntil(ctx, closeCondition) {
       continue;
     }
     let matched = false;
-    const blockCtx = { ...ctx, pos, blockCloseCondition: closeCondition };
+    const blockCtx = {
+      ...ctx,
+      pos,
+      blockRules,
+      blockCloseCondition: closeCondition,
+      excludedBlockNames: excluded
+    };
     for (const rule of blockRules) {
       if (canApplyBlockRule(rule, token)) {
         const result = rule.parse(blockCtx);
@@ -2619,44 +2637,7 @@ function consumeCloseTag2(ctx, pos) {
     closeConsumed++;
   return closeConsumed;
 }
-function mergeParagraphs(elements) {
-  const result = [];
-  let mergedElements = [];
-  for (const elem of elements) {
-    if (elem.element === "container" && elem.data && typeof elem.data === "object" && "type" in elem.data && elem.data.type === "paragraph") {
-      if (mergedElements.length > 0) {
-        mergedElements.push({ element: "line-break" });
-      }
-      if ("elements" in elem.data && Array.isArray(elem.data.elements)) {
-        mergedElements.push(...elem.data.elements);
-      }
-    } else {
-      if (mergedElements.length > 0) {
-        result.push({
-          element: "container",
-          data: {
-            type: "paragraph",
-            attributes: {},
-            elements: mergedElements
-          }
-        });
-        mergedElements = [];
-      }
-      result.push(elem);
-    }
-  }
-  if (mergedElements.length > 0) {
-    result.push({
-      element: "container",
-      data: {
-        type: "paragraph",
-        attributes: {},
-        elements: mergedElements
-      }
-    });
-  }
-  return result;
-}
+var EXCLUDED_BLOCKS = new Set(["collapsible"]);
 var collapsibleRule = {
   name: "collapsible",
   startTokens: ["BLOCK_OPEN"],
@@ -2720,18 +2701,16 @@ var collapsibleRule = {
         bodyElements = [];
       }
     } else {
-      const bodyCtx = {
-        ...ctx,
-        pos,
-        blockRules: ctx.blockRules.filter((r) => r.name !== "collapsible")
-      };
+      const bodyCtx = { ...ctx, pos };
       const closeCondition = (checkCtx) => {
         return isCollapsibleClose(checkCtx, checkCtx.pos);
       };
-      const bodyResult = parseBlocksUntil(bodyCtx, closeCondition);
+      const bodyResult = parseBlocksUntil(bodyCtx, closeCondition, {
+        excludedBlockNames: EXCLUDED_BLOCKS
+      });
       consumed += bodyResult.consumed;
       pos += bodyResult.consumed;
-      bodyElements = mergeParagraphs(bodyResult.elements);
+      bodyElements = bodyResult.elements;
     }
     if (!isCollapsibleClose(ctx, pos)) {
       ctx.diagnostics.push({

package/dist/index.js

@@ -591,7 +591,139 @@ var BLOCK_START_TOKENS = [
   "CLEAR_FLOAT_RIGHT"
 ];
 
+// packages/parser/src/parser/rules/utils.ts
+var SAFE_ATTRIBUTES = new Set([
+  "accept",
+  "align",
+  "alt",
+  "autocapitalize",
+  "autoplay",
+  "background",
+  "bgcolor",
+  "border",
+  "buffered",
+  "checked",
+  "cite",
+  "class",
+  "cols",
+  "colspan",
+  "contenteditable",
+  "controls",
+  "coords",
+  "datetime",
+  "decoding",
+  "default",
+  "dir",
+  "dirname",
+  "disabled",
+  "download",
+  "draggable",
+  "for",
+  "form",
+  "headers",
+  "height",
+  "hidden",
+  "high",
+  "href",
+  "hreflang",
+  "id",
+  "inputmode",
+  "ismap",
+  "itemprop",
+  "kind",
+  "label",
+  "lang",
+  "list",
+  "loop",
+  "low",
+  "max",
+  "maxlength",
+  "min",
+  "minlength",
+  "multiple",
+  "muted",
+  "name",
+  "optimum",
+  "pattern",
+  "placeholder",
+  "poster",
+  "preload",
+  "readonly",
+  "required",
+  "reversed",
+  "role",
+  "rows",
+  "rowspan",
+  "scope",
+  "selected",
+  "shape",
+  "size",
+  "sizes",
+  "span",
+  "spellcheck",
+  "src",
+  "srclang",
+  "srcset",
+  "start",
+  "step",
+  "style",
+  "tabindex",
+  "target",
+  "title",
+  "translate",
+  "type",
+  "usemap",
+  "value",
+  "width",
+  "wrap"
+]);
+function filterUnsafeAttributes(attrs) {
+  const result = {};
+  for (const [key, value] of Object.entries(attrs)) {
+    const lower = key.toLowerCase();
+    if (lower.startsWith("on"))
+      continue;
+    if (lower.startsWith("aria-") || lower.startsWith("data-")) {
+      result[key] = value;
+      continue;
+    }
+    if (!SAFE_ATTRIBUTES.has(lower))
+      continue;
+    if (lower === "id") {
+      result[key] = value.startsWith("u-") ? value : `u-${value}`;
+      continue;
+    }
+    result[key] = value;
+  }
+  return result;
+}
+function parseBlockName(ctx, startPos) {
+  let pos = startPos;
+  let consumed = 0;
+  const token = ctx.tokens[pos];
+  if (!token || token.type !== "TEXT" && token.type !== "IDENTIFIER") {
+    return null;
+  }
+  let name = token.value.toLowerCase();
+  consumed++;
+  pos++;
+  if (ctx.tokens[pos]?.type === "UNDERSCORE") {
+    name += "_";
+    consumed++;
+  }
+  return { name, consumed };
+}
+
 // packages/parser/src/parser/rules/inline/utils.ts
+function isExcludedBlockToken(ctx, tokenPos) {
+  if (!ctx.excludedBlockNames?.size)
+    return false;
+  const token = ctx.tokens[tokenPos];
+  if (token?.type !== "BLOCK_OPEN" && token?.type !== "BLOCK_END_OPEN")
+    return false;
+  const nameResult = parseBlockName(ctx, tokenPos + 1);
+  return nameResult !== null && ctx.excludedBlockNames.has(nameResult.name);
+}
 function canApplyInlineRule(rule, token) {
   if (rule.startTokens.length === 0) {
     return true;
@@ -683,7 +815,8 @@ function parseInlineUntil(ctx, endType) {
           isInvalidHeading = true;
         }
       }
-      const isBlockStart = nextMeaningfulToken && BLOCK_START_TOKENS.includes(nextMeaningfulToken.type) && nextMeaningfulToken.lineStart && !isOrphanCloseSpan && !isAnchorName && !isInvalidBlockOpen && !isInvalidHeading;
+      const isExcludedBlock = (nextMeaningfulToken?.type === "BLOCK_OPEN" || nextMeaningfulToken?.type === "BLOCK_END_OPEN") && isExcludedBlockToken(ctx, pos + lookAhead);
+      const isBlockStart = nextMeaningfulToken && BLOCK_START_TOKENS.includes(nextMeaningfulToken.type) && nextMeaningfulToken.lineStart && !isOrphanCloseSpan && !isAnchorName && !isInvalidBlockOpen && !isInvalidHeading && !isExcludedBlock;
       if (!nextMeaningfulToken || nextMeaningfulToken.type === "NEWLINE" || nextMeaningfulToken.type === "EOF" || isBlockStart) {
         if (isBlockStart && nodes.length > 0) {
           const nextPos = pos + lookAhead;
@@ -1021,129 +1154,6 @@ function buildListData(topLtype, list) {
   };
 }
 
-// packages/parser/src/parser/rules/utils.ts
-var SAFE_ATTRIBUTES = new Set([
-  "accept",
-  "align",
-  "alt",
-  "autocapitalize",
-  "autoplay",
-  "background",
-  "bgcolor",
-  "border",
-  "buffered",
-  "checked",
-  "cite",
-  "class",
-  "cols",
-  "colspan",
-  "contenteditable",
-  "controls",
-  "coords",
-  "datetime",
-  "decoding",
-  "default",
-  "dir",
-  "dirname",
-  "disabled",
-  "download",
-  "draggable",
-  "for",
-  "form",
-  "headers",
-  "height",
-  "hidden",
-  "high",
-  "href",
-  "hreflang",
-  "id",
-  "inputmode",
-  "ismap",
-  "itemprop",
-  "kind",
-  "label",
-  "lang",
-  "list",
-  "loop",
-  "low",
-  "max",
-  "maxlength",
-  "min",
-  "minlength",
-  "multiple",
-  "muted",
-  "name",
-  "optimum",
-  "pattern",
-  "placeholder",
-  "poster",
-  "preload",
-  "readonly",
-  "required",
-  "reversed",
-  "role",
-  "rows",
-  "rowspan",
-  "scope",
-  "selected",
-  "shape",
-  "size",
-  "sizes",
-  "span",
-  "spellcheck",
-  "src",
-  "srclang",
-  "srcset",
-  "start",
-  "step",
-  "style",
-  "tabindex",
-  "target",
-  "title",
-  "translate",
-  "type",
-  "usemap",
-  "value",
-  "width",
-  "wrap"
-]);
-function filterUnsafeAttributes(attrs) {
-  const result = {};
-  for (const [key, value] of Object.entries(attrs)) {
-    const lower = key.toLowerCase();
-    if (lower.startsWith("on"))
-      continue;
-    if (lower.startsWith("aria-") || lower.startsWith("data-")) {
-      result[key] = value;
-      continue;
-    }
-    if (!SAFE_ATTRIBUTES.has(lower))
-      continue;
-    if (lower === "id") {
-      result[key] = value.startsWith("u-") ? value : `u-${value}`;
-      continue;
-    }
-    result[key] = value;
-  }
-  return result;
-}
-function parseBlockName(ctx, startPos) {
-  let pos = startPos;
-  let consumed = 0;
-  const token = ctx.tokens[pos];
-  if (!token || token.type !== "TEXT" && token.type !== "IDENTIFIER") {
-    return null;
-  }
-  let name = token.value.toLowerCase();
-  consumed++;
-  pos++;
-  if (ctx.tokens[pos]?.type === "UNDERSCORE") {
-    name += "_";
-    consumed++;
-  }
-  return { name, consumed };
-}
-
 // packages/parser/src/parser/rules/block/utils.ts
 function canApplyBlockRule(rule, token) {
   if (rule.requiresLineStart && !token.lineStart) {
@@ -1154,11 +1164,13 @@ function canApplyBlockRule(rule, token) {
   }
   return rule.startTokens.includes(token.type);
 }
-function parseBlocksUntil(ctx, closeCondition) {
+function parseBlocksUntil(ctx, closeCondition, options) {
   const elements = [];
   let consumed = 0;
   let pos = ctx.pos;
-  const { blockRules, blockFallbackRule } = ctx;
+  const excluded = options?.excludedBlockNames;
+  const blockRules = excluded ? ctx.blockRules.filter((r) => !excluded.has(r.name)) : ctx.blockRules;
+  const { blockFallbackRule } = ctx;
   while (pos < ctx.tokens.length) {
     const token = ctx.tokens[pos];
     if (!token || token.type === "EOF") {
@@ -1179,7 +1191,13 @@ function parseBlocksUntil(ctx, closeCondition) {
       continue;
     }
     let matched = false;
-    const blockCtx = { ...ctx, pos, blockCloseCondition: closeCondition };
+    const blockCtx = {
+      ...ctx,
+      pos,
+      blockRules,
+      blockCloseCondition: closeCondition,
+      excludedBlockNames: excluded
+    };
     for (const rule of blockRules) {
       if (canApplyBlockRule(rule, token)) {
         const result = rule.parse(blockCtx);
@@ -2564,44 +2582,7 @@ function consumeCloseTag2(ctx, pos) {
     closeConsumed++;
   return closeConsumed;
 }
-function mergeParagraphs(elements) {
-  const result = [];
-  let mergedElements = [];
-  for (const elem of elements) {
-    if (elem.element === "container" && elem.data && typeof elem.data === "object" && "type" in elem.data && elem.data.type === "paragraph") {
-      if (mergedElements.length > 0) {
-        mergedElements.push({ element: "line-break" });
-      }
-      if ("elements" in elem.data && Array.isArray(elem.data.elements)) {
-        mergedElements.push(...elem.data.elements);
-      }
-    } else {
-      if (mergedElements.length > 0) {
-        result.push({
-          element: "container",
-          data: {
-            type: "paragraph",
-            attributes: {},
-            elements: mergedElements
-          }
-        });
-        mergedElements = [];
-      }
-      result.push(elem);
-    }
-  }
-  if (mergedElements.length > 0) {
-    result.push({
-      element: "container",
-      data: {
-        type: "paragraph",
-        attributes: {},
-        elements: mergedElements
-      }
-    });
-  }
-  return result;
-}
+var EXCLUDED_BLOCKS = new Set(["collapsible"]);
 var collapsibleRule = {
   name: "collapsible",
   startTokens: ["BLOCK_OPEN"],
@@ -2665,18 +2646,16 @@ var collapsibleRule = {
         bodyElements = [];
       }
     } else {
-      const bodyCtx = {
-        ...ctx,
-        pos,
-        blockRules: ctx.blockRules.filter((r) => r.name !== "collapsible")
-      };
+      const bodyCtx = { ...ctx, pos };
       const closeCondition = (checkCtx) => {
         return isCollapsibleClose(checkCtx, checkCtx.pos);
       };
-      const bodyResult = parseBlocksUntil(bodyCtx, closeCondition);
+      const bodyResult = parseBlocksUntil(bodyCtx, closeCondition, {
+        excludedBlockNames: EXCLUDED_BLOCKS
+      });
       consumed += bodyResult.consumed;
       pos += bodyResult.consumed;
-      bodyElements = mergeParagraphs(bodyResult.elements);
+      bodyElements = bodyResult.elements;
     }
     if (!isCollapsibleClose(ctx, pos)) {
       ctx.diagnostics.push({

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@wdprlib/parser",
-  "version": "2.0.9",
+  "version": "2.0.10",
   "description": "Parser for Wikidot markup",
   "keywords": [
     "ast",