jvr | d954dd2 | 2002-09-10 11:36:43 +0000 | [diff] [blame] | 1 | #! /usr/bin/env python |
| 2 | |
| 3 | |
| 4 | """This script builds the Lib/fontTools/ttLib/tables/otData.py file |
| 5 | from the OpenType HTML documentation. However, it depends on a slightly |
| 6 | patched version the the HTML, as there are some inconsistencies in the |
| 7 | markup and the naming of certain fields. See doco.diff for differences, |
| 8 | but this is probably against a slightly older version of the documentation |
| 9 | than what is currently online. The documentation was taken from this URL: |
| 10 | http://www.microsoft.com/typography/otspec/default.htm |
| 11 | """ |
| 12 | |
| 13 | |
| 14 | from sgmllib import SGMLParser |
| 15 | |
| 16 | |
| 17 | class HTMLParser(SGMLParser): |
| 18 | |
| 19 | def __init__(self): |
| 20 | SGMLParser.__init__(self) |
| 21 | self.data = None |
| 22 | self.currenttable = None |
| 23 | self.lastcaption = None |
| 24 | |
| 25 | def handle_data(self, data): |
| 26 | if self.data is not None: |
| 27 | self.data.append(data) |
| 28 | |
| 29 | def start_i(self, attrs): |
| 30 | if self.currenttable is None: |
| 31 | self.data = [] |
| 32 | def end_i(self): |
| 33 | if self.currenttable is None: |
| 34 | self.lastcaption = " ".join(self.data) |
| 35 | self.data = None |
| 36 | |
| 37 | def start_b(self, attrs): |
| 38 | if self.currenttable is None: |
| 39 | self.data = [] |
| 40 | def end_b(self): |
| 41 | if self.currenttable is None: |
| 42 | self.lastcaption = " ".join(self.data) |
| 43 | self.data = None |
| 44 | |
| 45 | def start_table(self, attrs): |
| 46 | attrs = dict(attrs) |
| 47 | if attrs.get('width') in ('455', '460'): |
| 48 | #print "---", attrs |
| 49 | self.currenttable = [] |
| 50 | else: |
| 51 | self.currenttable = None |
| 52 | def end_table(self): |
| 53 | if self.currenttable is not None and self.lastcaption is not None: |
| 54 | if self.currenttable[0] == ['Type', 'Name', 'Description'] or \ |
| 55 | self.currenttable[0] == ['Value', 'Type', 'Description']: |
| 56 | caption = self.lastcaption.split() |
| 57 | name = caption[0] |
| 58 | if name == "LookupType" or name == "LookupFlag": |
| 59 | self.currenttable = None |
| 60 | return |
| 61 | elif name == "Device": |
| 62 | if "Tables" in caption: |
| 63 | # XXX skip this one |
| 64 | self.currenttable = None |
| 65 | return |
| 66 | buildTable(name, self.currenttable[1:], self.lastcaption) |
| 67 | self.currenttable = None |
| 68 | |
| 69 | def start_tr(self, attrs): |
| 70 | if self.currenttable is not None: |
| 71 | self.currenttable.append([]) |
| 72 | def end_tr(self): |
| 73 | pass |
| 74 | |
| 75 | def start_td(self, attrs): |
| 76 | self.data = [] |
| 77 | def end_td(self): |
| 78 | if self.currenttable is not None and self.data is not None: |
| 79 | self.currenttable[-1].append(" ".join(self.data)) |
| 80 | self.data = None |
| 81 | |
| 82 | |
| 83 | globalDups = {} |
| 84 | localDups = {} |
| 85 | not3 = [] |
| 86 | |
| 87 | def buildTable(name, table, caption): |
| 88 | if globalDups.has_key(name): |
| 89 | globalDups[name].append(caption) |
| 90 | else: |
| 91 | globalDups[name] = [caption] |
| 92 | print "\t(%s, [" % repr(name) |
| 93 | allFields = {} |
| 94 | for row in table: |
| 95 | row = [" ".join(x.split()) for x in row] |
| 96 | if len(row) <> 3: |
| 97 | not3.append(row) |
| 98 | row = makeRow(row) |
| 99 | fieldName = row[1] |
| 100 | if allFields.has_key(fieldName): |
| 101 | key = (name, fieldName) |
| 102 | localDups[key] = 1 |
| 103 | allFields[fieldName] = 1 |
| 104 | print "\t\t%s," % (tuple(row),) |
| 105 | print "\t])," |
| 106 | print |
| 107 | |
| 108 | |
| 109 | def makeRow(rawRow): |
| 110 | tp, name = rawRow[:2] |
| 111 | name = name.strip() |
| 112 | rest = tuple(rawRow[2:]) |
| 113 | if '[' in name: |
| 114 | name, repeat = name.split("[") |
| 115 | name = name.strip() |
| 116 | assert repeat[-1] == "]" |
| 117 | repeat = repeat[:-1].split() |
| 118 | if repeat[1:]: |
| 119 | repeatOffset = int("".join(repeat[1:])) |
| 120 | else: |
| 121 | repeatOffset = 0 |
| 122 | if not repeat: |
| 123 | repeat = "" |
| 124 | else: |
| 125 | repeat = repeat[0] |
| 126 | else: |
| 127 | repeat = None |
| 128 | repeatOffset = None |
| 129 | row = (tp, name, repeat, repeatOffset) + rest |
| 130 | return row |
| 131 | |
| 132 | |
| 133 | if __name__ == "__main__": |
| 134 | import sys, os |
| 135 | if "-" not in sys.argv: |
| 136 | sys.stdout = open("otData.py", "w") |
| 137 | print "otData = [" |
| 138 | for file in ["chapter2.htm", "gpos.htm", "gsub.htm", "gdef.htm", "base.htm", "jstf.htm"]: |
| 139 | name = os.path.splitext(file)[0] |
| 140 | if name == "chapter2": |
| 141 | name = "common" |
| 142 | print |
| 143 | print "\t#" |
| 144 | print "\t# %s (generated from %s)" % (name, file) |
| 145 | print "\t#" |
| 146 | print |
| 147 | p = HTMLParser() |
| 148 | p.feed(open(file).read()) |
| 149 | p.close() |
| 150 | print "]" |
| 151 | print |
| 152 | for k, v in globalDups.items(): |
| 153 | if len(v) > 1: |
| 154 | print "# XXX duplicate table name:", k, v |
| 155 | for (name, fieldName), v in localDups.items(): |
| 156 | print "# XXX duplicate field name '%s' in table '%s'" % (fieldName, name) |
| 157 | for n in not3: |
| 158 | print "#XXX", not3 |