feat(ooxast-unified): add lists

Author: tefkah

libs/jast/jast-util-to-texast/src/lib/handlers/graphic.ts

@@ -1,5 +1,5 @@
 import { Graphic, List } from 'jast-types'
-import { all } from 'jast-util-to-texast'
+import { all } from '../all'
 import { J } from '../types'
 import { wrapCommandArg } from '../util/wrap-command-arg'
 

libs/jast/jast-util-to-texast/src/lib/handlers/sec.ts

@@ -43,7 +43,7 @@ export function sec(j: J, node: Parents) {
         type: 'commandArg',
         optional: false,
         // TODO: [jast-util-to-texast] Figure out how to properly nest environments and commands
-        // @ts-expect-error
+        // @ts-expect-error theres something wrong with this nesting
         children: all(j, titleElement),
       },
     ],

libs/ooxast/ooxast-util-to-unified-latex/src/lib/handlers/body.ts

@@ -1,10 +1,202 @@
 // based on https://github.com/syntax-tree/hast-util-to-mdast/blob/main/lib/handlers/em
 
-import { H, Body, Handle } from '../types'
+import { H, Body, Handle, P, UnifiedLatexNode } from '../types'
 import { all } from '../all'
+import { getPStyle } from '../util/get-pstyle'
+import { getListInfo } from '../util/get-listinfo'
+import { one } from '../one'
+import { Element } from 'xast-util-to-string/lib'
+import {
+  Argument,
+  Environment,
+  Macro,
+  Verb,
+  Whitespace,
+} from '@unified-latex/unified-latex-types'
+import { WhiteSpace } from 'nlcst'
+import { arg, m, SP } from '@unified-latex/unified-latex-builder'
+import { PB } from '../util/PB'
+import { updateRenderInfo } from '@unified-latex/unified-latex-util-render-info'
+
+const isP = (node: Element): node is P =>
+  node.type === 'element' && node.name === 'w:p'
 
 export const body: Handle = (h: H, body: Body) => {
   const bod = all(h, body)
 
-  return bod
+  const processedBody = body.children.reduce((acc, child, index) => {
+    if (child.type !== 'element') return acc
+
+    if (!isP(child)) {
+      acc.push(...makeOne(h, child, body))
+      return acc
+    }
+
+    const isListItem = getPStyle(child) === 'ListParagraph'
+
+    if (!isListItem) {
+      acc.push(...makeOne(h, child, body))
+      return acc
+    }
+
+    const { ilvl, numId } = getListInfo(child) ?? {}
+
+    if (ilvl == null || numId == null) {
+      acc.push(...makeOne(h, child, body))
+      return acc
+    }
+
+    const prevChild = body.children[index - 1]
+
+    const isPrevListItem =
+      prevChild && isP(prevChild) && getPStyle(prevChild) === 'ListParagraph'
+
+    const { ilvl: prevIlvl, numId: prevNumId } =
+      (isPrevListItem && getListInfo(prevChild)) || {}
+
+    const listItem = makeItem(h, child)
+
+    // there's no previous list item, we need to create a new environment
+    if (!isPrevListItem || prevIlvl == null || prevNumId == null) {
+      const env: Environment = {
+        type: 'environment',
+        env: 'enumerate',
+        content: listItem,
+        _renderInfo: {
+          ilvl,
+          numId,
+        },
+      }
+
+      acc.push(env)
+      return acc
+    }
+
+    // there is a previous list item, and it's the same level, so we need to add it to the previous environment
+    if (isPrevListItem && ilvl === prevIlvl && numId === prevNumId) {
+      const mainEnv = acc[acc.length - 1] as Environment
+
+      if (mainEnv.env !== 'enumerate') {
+        throw new Error('prevEnv.env !== enumerate')
+      }
+
+      const embeddedEnvs = findEmbeddedEnvs(mainEnv)
+      const lastEnv = embeddedEnvs[embeddedEnvs.length - 1]
+
+      lastEnv.content.push(...listItem)
+      return acc
+    }
+
+    // there is a previous list item, and it's a lower level, so we need to add a new environment inside the previous environment
+    if (isPrevListItem && ilvl > prevIlvl) {
+      const mainEnv = acc[acc.length - 1] as Environment
+      const embeddedEnvs = findEmbeddedEnvs(mainEnv)
+      const lastEnv = embeddedEnvs[embeddedEnvs.length - 1]
+      const env: Environment = {
+        type: 'environment',
+        env: 'enumerate',
+        content: listItem,
+        _renderInfo: {
+          ilvl,
+          numId,
+        },
+      }
+      lastEnv.content.push(env)
+      return acc
+    }
+
+    // there is a previous list item, and it's a higher level, so we need to close the previous environment and create a new one
+    if (isPrevListItem && ilvl < prevIlvl) {
+      const mainEnv = acc[acc.length - 1] as Environment
+      const embeddedEnvs = findEmbeddedEnvs(mainEnv)
+      const toBeEmbeddedEnv =
+        embeddedEnvs[embeddedEnvs.length - (prevIlvl - ilvl) - 1]
+
+      const env: Environment = {
+        type: 'environment',
+        env: 'enumerate',
+        content: listItem,
+        _renderInfo: {
+          ilvl,
+          numId,
+        },
+      }
+
+      if (!toBeEmbeddedEnv) {
+        acc.push(env)
+        return acc
+      }
+
+      if (toBeEmbeddedEnv._renderInfo?.ilvl !== ilvl) {
+        toBeEmbeddedEnv.content.push(env)
+        return acc
+      }
+
+      const item = makeItem(h, child)
+
+      toBeEmbeddedEnv.content.push(...item)
+      return acc
+    }
+
+    // if we are immediately starting a new, but different list after a previous list, we need to close the previous list and start a new one
+    if (isPrevListItem && ilvl === prevIlvl && numId !== prevNumId) {
+      const mainEnv = acc[acc.length - 1] as Environment
+      const embeddedEnvs = findEmbeddedEnvs(mainEnv)
+      const lastEnv = embeddedEnvs[embeddedEnvs.length - 1]
+      const env: Environment = {
+        type: 'environment',
+        env: 'enumerate',
+        content: listItem,
+        _renderInfo: {
+          ilvl,
+          numId,
+        },
+      }
+      lastEnv.content.push(env)
+      return acc
+    }
+
+    return acc
+  }, [] as UnifiedLatexNode[])
+  return processedBody
+}
+
+function makeItem(
+  h: H,
+  item: P
+): [typeof PB, Macro, typeof SP, ...UnifiedLatexNode[], typeof PB] {
+  const mIte = m('item')
+  updateRenderInfo(mIte, {
+    hangingIndent: true,
+    inParMode: true,
+  })
+
+  return [PB, mIte, SP, ...all(h, item), PB]
+}
+
+function findEmbeddedEnvs(envs: Environment[] | Environment): Environment[] {
+  envs = Array.isArray(envs) ? envs : [envs]
+
+  if (envs.length === 0) return []
+  const env = envs[envs.length - 1]
+  if (env.content.length === 0) return envs
+
+  const last = env.content[env.content.length - 1]
+
+  if (last.type !== 'environment') return envs
+
+  return findEmbeddedEnvs([...envs, last])
+}
+
+function makeOne(h: H, node: Element, body: Body): UnifiedLatexNode[] {
+  const res = one(h, node, body)
+  if (!res) {
+    return []
+  }
+
+  if (Array.isArray(res)) {
+    return res
+  }
+
+  return [res]
 }

libs/ooxast/ooxast-util-to-unified-latex/src/lib/handlers/p.spec.ts

@@ -1,5 +1,7 @@
 import { x } from 'xastscript'
 import { toUnifiedLatex } from '../ooxast-util-to-unified-latex'
+import { s, SP } from '@unified-latex/unified-latex-builder'
+
 describe('p', () => {
   it('should do something', () => {
     const basicp = x('p', { id: 'ayy' }, [
@@ -9,6 +11,6 @@ describe('p', () => {
         x('w:t', {}, [{ type: 'text', value: 'lmao' }]),
       ]),
     ])
-    expect(toUnifiedLatex(basicp)).toEqual(x('p', [{ type: 'text', value: 'lmao' }]))
+    expect(toUnifiedLatex(basicp)).toEqual([SP])
   })
 })

libs/ooxast/ooxast-util-to-unified-latex/src/lib/handlers/root.ts

@@ -5,25 +5,9 @@ import { H, Node, Root } from '../types'
 
 export function root(h: H, node: Root) {
   //  if (!article) {
-  const children: any[] = [
-    {
-      type: 'instruction',
-      name: 'xml',
-      value: 'version="1.0" encoding="utf-8"',
-    },
-    {
-      type: 'doctype',
-      name: 'article',
-      public:
-        '-//NLM//DTD JATS (Z39.96) Journal Publishing DTD v1.2 20190208//EN',
-      system:
-        'https://jats.nlm.nih.gov/publishing/1.2/JATS-journalpublishing1.dtd',
-    },
-  ]
 
-  const content = all(j, node)
-  content.unshift(...children)
-  return j(node, 'root', content)
+  const content = all(h, node)
+  return h(node, 'root', content)
   //  }
 
   // return j(article, 'root', [

libs/ooxast/ooxast-util-to-unified-latex/src/lib/ooxast-util-to-unified-latex.spec.ts

@@ -1,11 +1,12 @@
-import { P } from 'ooxast'
+import { Document, P, Root } from 'ooxast'
 import { Macro } from '@unified-latex/unified-latex-types'
 import { p } from './handlers/p'
 import { toUnifiedLatex } from './ooxast-util-to-unified-latex'
 import { toString } from '@unified-latex/unified-latex-util-to-string'
 import { m } from '@unified-latex/unified-latex-builder'
 import { updateRenderInfo } from '@unified-latex/unified-latex-util-render-info'
 import { PB } from './util/PB'
+import listTree from '../lib/test/list.json'
 
 // test whether a ooxast p node with style 'Heading1' is converted to a macro with name 'section'
 describe('ooxast-util-to-unified-latex', () => {
@@ -71,4 +72,37 @@ describe('ooxast-util-to-unified-latex', () => {
     )
     expect(toUnifiedLatex(ooxastParagraph)).toEqual(res)
   })
+
+  it('should convert a file with lists to nested enum envs', () => {
+    const res = toUnifiedLatex(listTree as Root)
+    console.dir(res, { depth: null })
+    expect(toString(res)).toEqual(`\\begin{enumerate}··
+    \\item Number one···
+    \\item Two···
+    \\item Three·
+    \\begin{enumerate}··
+            \\item Indent··
+    \\end{enumerate}
+\\end{enumerate}···
+\\begin{enumerate}··
+    \\item Unordered···
+    \\item List···
+    \\item Items·
+    \\begin{enumerate}··
+            \\item Indentatio··
+    \\end{enumerate}
+\\end{enumerate}···
+\\begin{enumerate}··
+    \\item Continuing···
+    \\item Previous···
+    \\item List·
+    \\begin{enumerate}··
+            \\item Indentation··
+    \\end{enumerate}·
+    \\item Going down a step again·
+    \\begin{enumerate}··
+            \\item Immediately starting other list··
+    \\end{enumerate}
+\\end{enumerate}`)
+  })
 })