add better support for tables

Author: tefkah

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

@@ -12,6 +12,9 @@ const visit = origVisit as any
 export function table(j: J, table: Table) {
   let columns: string[] = []
   let hasCols = false
+
+  let numberOfRows = 0
+
   // tables can be def'd in terms of cols or rows
   // TODO: [jast-util-to-texast] Store information about column alignment in tabular
   visit(
@@ -20,15 +23,28 @@ export function table(j: J, table: Table) {
     (node: Col | Tr) => {
       if (node.name === 'col') {
         hasCols = true
-        columns.push('c')
+        columns.push('l')
+
+        // TODO: [jast-util-to-texast] Check whether col how multirow elements
+        numberOfRows = node.children.length
+
         return
       }
 
       if (hasCols) return
+
+      numberOfRows += 1
+
       let tempCols: string[] = []
 
       node?.children?.forEach((child) => {
-        isElement(child) && child.name === 'td' && tempCols.push('c')
+        if (!isElement(child)) return
+
+        if (child.name !== 'td') return
+
+        for (let i = 0; i < (child?.attributes?.colspan ?? 1); i++) {
+          tempCols.push('l')
+        }
       })
       // Just make the table as wide as it needs to be, overfull tables
       // error out while  underfull ones are fine
@@ -39,12 +55,16 @@ export function table(j: J, table: Table) {
   )
 
   const colAlignment = columns.join(` ${j.columnSeparator ? '|' : ''} `)
+
   const colAlignArg: CommandArg = {
     type: 'commandArg',
     children: [{ type: 'text', value: colAlignment }],
   }
 
+  j.numberOfRows = numberOfRows
   const contents = all(j, table)
+  j.rowNumber = 0
+  j.numberOfRows = 0
   contents.unshift(colAlignArg)
 
   return { type: 'environment', name: 'tabular', children: contents }

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

@@ -1,9 +1,11 @@
 // based on https://github.com/syntax-tree/hast-util-to-mdast/blob/main/lib/handlers/em
 
 import { Td } from 'jast-types'
+import { TableCell } from 'texast'
 import { all } from '../all'
 import { J, Node } from '../types'
 
 export function td(j: J, node: Td) {
-  return j(node, 'tableCell', all(j, node))
+  const { colspan, style, valign } = node.attributes
+  return j(node, 'tableCell', { span: colspan }, all(j, node)) as TableCell
 }

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

@@ -1,11 +1,12 @@
 // based on https://github.com/syntax-tree/hast-util-to-mdast/blob/main/lib/handlers/em
 
 import { Tr } from 'jast-types'
-import { AlignmentTab, Linebreak } from 'texast'
+import { AlignmentTab, Command, Linebreak, Text } from 'texast'
 import { all } from '../all'
 import { J, Node } from '../types'
 
 export function tr(j: J, node: Tr) {
+  j.rowNumber += 1
   const contentsNoTabs = all(j, node)
   // too many alignment tabs
   const contents = contentsNoTabs
@@ -17,5 +18,47 @@ export function tr(j: J, node: Tr) {
     .slice(1)
 
   contents.push({ type: 'linebreak', value: '\\' } as Linebreak)
+  if (
+    node.children.every(
+      (child) =>
+        child.type === 'element' &&
+        child.name === 'td' &&
+        child.attributes.style?.includes('border-bottom')
+    )
+  ) {
+    const commandName = j.booktabs
+      ? j.rowNumber === j.numberOfRows
+        ? 'bottomrule'
+        : 'midrule'
+      : 'hline'
+    contents.push(
+      ...[
+        { type: 'text', value: ' ' } as Text,
+        {
+          type: 'command',
+          name: commandName,
+          children: [],
+        } as Command,
+      ]
+    )
+  }
+
+  if (
+    j.rowNumber === 1 &&
+    node.children.every(
+      (child) =>
+        child.type === 'element' &&
+        child.name === 'td' &&
+        child.attributes.style?.includes('border-top')
+    )
+  ) {
+    const commandName = j.booktabs ? 'toprule' : 'hline'
+    contents.unshift({
+      type: 'command',
+      name: commandName,
+      children: [],
+    } as Command)
+  }
+
   return j(node, 'tableRow', contents)
 }

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

@@ -91,6 +91,9 @@ export function toTexast(
       bibname: options.bibname || 'bibliography',
       columnSeparator: !!options.columnSeparator,
       footnotes: [],
+      booktabs: options.booktabs ?? true,
+      rowNumber: 0,
+      numberOfRows: 0,
       citationAnalyzer:
         options.citationAnalyzer || ((node: Node) => 'autocite'),
     } as Context

libs/jast/jast-util-to-texast/src/lib/types.ts

@@ -37,6 +37,7 @@ export interface Options {
   italics?: 'emph' | 'textit'
   bibname?: string
   columnSeparator?: boolean
+  booktabs?: boolean
   documentClass?: {
     options?: string[]
     name: string
@@ -63,6 +64,7 @@ export interface Context {
   document: boolean | undefined
   checked: string
   unchecked: string
+  booktabs: boolean
   quotes: Array<string>
   italics: string
   documentClass: {
@@ -73,6 +75,8 @@ export interface Context {
   bibname: string
   columnSeparator: boolean
   footnotes: TexastContent[]
+  rowNumber: number
+  numberOfRows: number
   citationAnalyzer: (node: Node) => string
 }
 

libs/ooxast/ooxast-util-citations/src/lib/ooxast-util-citations.ts

@@ -1,9 +1,9 @@
-import { visit } from 'unist-util-visit'
+import { visitParents } from 'unist-util-visit-parents'
 import { P, R, T, Text, Root } from 'ooxast'
 import { getPStyle } from 'ooxast-util-get-style'
 import { parseTextCite } from 'parse-text-cite'
 import { Node } from 'unist'
-import { convertElement } from 'xast-util-is-element'
+import { convertElement, isElement } from 'xast-util-is-element'
 import { select } from 'xast-util-select'
 import { toString } from 'xast-util-to-string'
 import { select as unistSelect } from 'unist-util-select'
@@ -37,8 +37,16 @@ export function findCitations(
 
   let references = false
   let citationCounter = 1
-  visit(tree, isP, (p: P) => {
-    if (references) return
+  visitParents(tree, isP, (p: P, ancestors: Node[]) => {
+    if (
+      ancestors.some((parent) => isElement(parent) && /tbl/.test(parent.name))
+    ) {
+      return
+    }
+
+    if (references) {
+      return
+    }
 
     if (getPStyle(p)?.toLowerCase()?.includes('heading')) {
       if (

libs/ooxast/ooxast-util-to-jast/src/lib/handlers/index.ts

@@ -12,6 +12,7 @@ import { document } from './document'
 import { r } from './r'
 import { tr } from './tr'
 import { drawing } from './drawing'
+import { tc } from './tc'
 
 export const handlers = {
   p,
@@ -26,6 +27,7 @@ export const handlers = {
   instrText: citation,
   tbl,
   tr,
+  tc,
   article: ignore,
   xml: ignore,
   instruction: ignore,

libs/ooxast/ooxast-util-to-jast/src/lib/handlers/tc.ts

@@ -4,7 +4,56 @@ import { Tc } from 'ooxast'
 import { Td } from 'jast-types'
 import { getPr } from 'ooxast-util-properties'
 
+const createStyle = (
+  name: string,
+  style: Record<string, boolean | string | Record<string, string>>
+) =>
+  Object.entries(style)
+    ?.map(
+      ([key, val]) =>
+        `${name}${key?.at(0)?.toUpperCase()}${key?.slice(1)}: ${
+          val?.val ? val.val : `${val}`
+        }`
+    )
+    .join('; ')
+
 export function tc(j: J, node: Tc): Td {
   const properties = getPr(node)
-  return j(node, 'td', all(j, node))
+
+  const { tcBorders, gridSpan, vAlign, shd } = properties
+
+  const shdStyle =
+    shd && typeof shd === 'object'
+      ? `background-color: ${shd.fill}; color: ${shd.color}`
+      : ''
+
+  const borderStyle =
+    tcBorders && typeof tcBorders === 'object'
+      ? Object.entries(tcBorders)
+          ?.map(
+            ([border, values]) =>
+              `border-${border}: ${values.sz}px ${
+                values.val?.includes('dash') ? 'dashed' : 'solid'
+              } #${values.color}`
+          )
+          ?.join('; ')
+      : ''
+
+  // const borderStyle =
+  //   tcBorders && typeof tcBorders === 'object'
+  //     ? createStyle('border', tcBorders)
+  //     : ''
+
+  const style = [shdStyle, borderStyle].join('; ')
+
+  const jatsProps = {
+    ...(gridSpan && gridSpan.val && +gridSpan.val > 1
+      ? { colspan: gridSpan.val }
+      : {}),
+    ...(shdStyle ?? borderStyle ? { style } : {}),
+
+    ...(vAlign && vAlign?.val !== 'top' ? { valign: vAlign.val } : {}),
+  }
+
+  return j(node, 'td', jatsProps, all(j, node)) as Td
 }

libs/ooxast/ooxast-util-to-jast/src/lib/handlers/tr.ts

@@ -4,16 +4,16 @@ import { Td } from 'jast-types'
 import { all } from '../all'
 
 export function tr(j: J, tr: Row) {
-  const rowcontents = all(j, tr)
-  const celldRowContents: Td[] = []
-  for (const cell of rowcontents) {
-    celldRowContents.push({
-      type: 'element',
-      name: 'td',
-      attributes: {},
-      // @ts-expect-error
-      children: cell?.children || [cell],
-    })
-  }
-  return j(tr, 'tr', celldRowContents)
+  const rowContents = all(j, tr)
+  // const celldRowContents: Td[] = []
+  // for (const cell of rowcontents) {
+  //   celldRowContents.push({
+  //     type: 'element',
+  //     name: 'td',
+  //     attributes: {},
+  //     // @ts-expect-error
+  //     children: cell?.children || [cell],
+  //   })
+  // }
+  return j(tr, 'tr', rowContents)
 }

libs/ooxast/ooxast/src/lib/ooxml/wordprocessingml/main.ts

@@ -3979,7 +3979,7 @@ export type TcPr = TcPrInner & {
   children: RequiredMap<TcPrMap>[]
 }
 
-export interface TcPrMap {
+export interface TcPrMap extends TcPrBase {
   tcPrChange?: TcPrChange
 }