table.go

     1  // Copyright 2023 The Go Authors. All rights reserved.
     2  // Use of this source code is governed by a BSD-style
     3  // license that can be found in the LICENSE file.
     4  
     5  package markdown
     6  
     7  import (
     8  	"bytes"
     9  	"strings"
    10  )
    11  
    12  type tableTrimmed string
    13  
    14  func isTableSpace(c byte) bool {
    15  	return c == ' ' || c == '\t' || c == '\v' || c == '\f'
    16  }
    17  
    18  func tableTrimSpace(s string) string {
    19  	i := 0
    20  	for i < len(s) && isTableSpace(s[i]) {
    21  		i++
    22  	}
    23  	j := len(s)
    24  	for j > i && isTableSpace(s[j-1]) {
    25  		j--
    26  	}
    27  	return s[i:j]
    28  }
    29  
    30  func tableTrimOuter(row string) tableTrimmed {
    31  	row = tableTrimSpace(row)
    32  	if len(row) > 0 && row[0] == '|' {
    33  		row = row[1:]
    34  	}
    35  	if len(row) > 0 && row[len(row)-1] == '|' {
    36  		row = row[:len(row)-1]
    37  	}
    38  	return tableTrimmed(row)
    39  }
    40  
    41  func isTableStart(hdr1, delim1 string) bool {
    42  	// Scan potential delimiter string, counting columns.
    43  	// This happens on every line of text,
    44  	// so make it relatively quick - nothing expensive.
    45  	col := 0
    46  	delim := tableTrimOuter(delim1)
    47  	i := 0
    48  	for ; ; col++ {
    49  		for i < len(delim) && isTableSpace(delim[i]) {
    50  			i++
    51  		}
    52  		if i >= len(delim) {
    53  			break
    54  		}
    55  		if i < len(delim) && delim[i] == ':' {
    56  			i++
    57  		}
    58  		if i >= len(delim) || delim[i] != '-' {
    59  			return false
    60  		}
    61  		i++
    62  		for i < len(delim) && delim[i] == '-' {
    63  			i++
    64  		}
    65  		if i < len(delim) && delim[i] == ':' {
    66  			i++
    67  		}
    68  		for i < len(delim) && isTableSpace(delim[i]) {
    69  			i++
    70  		}
    71  		if i < len(delim) && delim[i] == '|' {
    72  			i++
    73  		}
    74  	}
    75  
    76  	if strings.TrimSpace(hdr1) == "|" {
    77  		// https://github.com/github/cmark-gfm/pull/127 and
    78  		// https://github.com/github/cmark-gfm/pull/128
    79  		// fixed a buffer overread by rejecting | by itself as a table line.
    80  		// That seems to violate the spec, but we will play along.
    81  		return false
    82  	}
    83  
    84  	return col == tableCount(tableTrimOuter(hdr1))
    85  }
    86  
    87  func tableCount(row tableTrimmed) int {
    88  	col := 1
    89  	prev := byte(0)
    90  	for i := 0; i < len(row); i++ {
    91  		c := row[i]
    92  		if c == '|' && prev != '\\' {
    93  			col++
    94  		}
    95  		prev = c
    96  	}
    97  	return col
    98  }
    99  
   100  type tableBuilder struct {
   101  	hdr   tableTrimmed
   102  	delim tableTrimmed
   103  	rows  []tableTrimmed
   104  }
   105  
   106  func (b *tableBuilder) start(hdr, delim string) {
   107  	b.hdr = tableTrimOuter(hdr)
   108  	b.delim = tableTrimOuter(delim)
   109  }
   110  
   111  func (b *tableBuilder) addRow(row string) {
   112  	b.rows = append(b.rows, tableTrimOuter(row))
   113  }
   114  
   115  type Table struct {
   116  	Position
   117  	Header []*Text
   118  	Align  []string // 'l', 'c', 'r' for left, center, right; 0 for unset
   119  	Rows   [][]*Text
   120  }
   121  
   122  func (t *Table) PrintHTML(buf *bytes.Buffer) {
   123  	buf.WriteString("<table>\n")
   124  	buf.WriteString("<thead>\n")
   125  	buf.WriteString("<tr>\n")
   126  	for i, hdr := range t.Header {
   127  		buf.WriteString("<th")
   128  		if t.Align[i] != "" {
   129  			buf.WriteString(" align=\"")
   130  			buf.WriteString(t.Align[i])
   131  			buf.WriteString("\"")
   132  		}
   133  		buf.WriteString(">")
   134  		hdr.PrintHTML(buf)
   135  		buf.WriteString("</th>\n")
   136  	}
   137  	buf.WriteString("</tr>\n")
   138  	buf.WriteString("</thead>\n")
   139  	if len(t.Rows) > 0 {
   140  		buf.WriteString("<tbody>\n")
   141  		for _, row := range t.Rows {
   142  			buf.WriteString("<tr>\n")
   143  			for i, cell := range row {
   144  				buf.WriteString("<td")
   145  				if i < len(t.Align) && t.Align[i] != "" {
   146  					buf.WriteString(" align=\"")
   147  					buf.WriteString(t.Align[i])
   148  					buf.WriteString("\"")
   149  				}
   150  				buf.WriteString(">")
   151  				cell.PrintHTML(buf)
   152  				buf.WriteString("</td>\n")
   153  			}
   154  			buf.WriteString("</tr>\n")
   155  		}
   156  		buf.WriteString("</tbody>\n")
   157  	}
   158  	buf.WriteString("</table>\n")
   159  }
   160  
   161  func (t *Table) printMarkdown(buf *bytes.Buffer, s mdState) {
   162  }
   163  
   164  func (b *tableBuilder) build(p buildState) Block {
   165  	pos := p.pos()
   166  	pos.StartLine-- // builder does not count header
   167  	pos.EndLine = pos.StartLine + 1 + len(b.rows)
   168  	t := &Table{
   169  		Position: pos,
   170  	}
   171  	width := tableCount(b.hdr)
   172  	t.Header = b.parseRow(p, b.hdr, pos.StartLine, width)
   173  	t.Align = b.parseAlign(b.delim, width)
   174  	t.Rows = make([][]*Text, len(b.rows))
   175  	for i, row := range b.rows {
   176  		t.Rows[i] = b.parseRow(p, row, pos.StartLine+2+i, width)
   177  	}
   178  	return t
   179  }
   180  
   181  func (b *tableBuilder) parseRow(p buildState, row tableTrimmed, line int, width int) []*Text {
   182  	out := make([]*Text, 0, width)
   183  	pos := Position{StartLine: line, EndLine: line}
   184  	start := 0
   185  	unesc := nop
   186  	for i := 0; i < len(row); i++ {
   187  		c := row[i]
   188  		if c == '\\' && i+1 < len(row) && row[i+1] == '|' {
   189  			unesc = tableUnescape
   190  			i++
   191  			continue
   192  		}
   193  		if c == '|' {
   194  			out = append(out, p.newText(pos, unesc(strings.Trim(string(row[start:i]), " \t\v\f"))))
   195  			if len(out) == width {
   196  				// Extra cells are discarded!
   197  				return out
   198  			}
   199  			start = i + 1
   200  			unesc = nop
   201  		}
   202  	}
   203  	out = append(out, p.newText(pos, unesc(strings.Trim(string(row[start:]), " \t\v\f"))))
   204  	for len(out) < width {
   205  		// Missing cells are considered empty.
   206  		out = append(out, p.newText(pos, ""))
   207  	}
   208  	return out
   209  }
   210  
   211  func nop(text string) string {
   212  	return text
   213  }
   214  
   215  func tableUnescape(text string) string {
   216  	out := make([]byte, 0, len(text))
   217  	for i := 0; i < len(text); i++ {
   218  		c := text[i]
   219  		if c == '\\' && i+1 < len(text) && text[i+1] == '|' {
   220  			i++
   221  			c = '|'
   222  		}
   223  		out = append(out, c)
   224  	}
   225  	return string(out)
   226  }
   227  
   228  func (b *tableBuilder) parseAlign(delim tableTrimmed, n int) []string {
   229  	align := make([]string, 0, tableCount(delim))
   230  	start := 0
   231  	for i := 0; i < len(delim); i++ {
   232  		if delim[i] == '|' {
   233  			align = append(align, tableAlign(string(delim[start:i])))
   234  			start = i + 1
   235  		}
   236  	}
   237  	align = append(align, tableAlign(string(delim[start:])))
   238  	return align
   239  }
   240  
   241  func tableAlign(cell string) string {
   242  	cell = tableTrimSpace(cell)
   243  	l := cell[0] == ':'
   244  	r := cell[len(cell)-1] == ':'
   245  	switch {
   246  	case l && r:
   247  		return "center"
   248  	case l:
   249  		return "left"
   250  	case r:
   251  		return "right"
   252  	}
   253  	return ""
   254  }
   255
View as plain text