| 1 | #!/usr/bin/env python
|
| 2 | from __future__ import print_function
|
| 3 | """
|
| 4 | csv2html.py
|
| 5 |
|
| 6 | Usage:
|
| 7 | csv2html.py foo.csv
|
| 8 |
|
| 9 | Attempts to read foo_schema.csv. If not it assumes everything is a string.
|
| 10 |
|
| 11 | Things it handles:
|
| 12 |
|
| 13 | - table-sort.js integration <colgroup>
|
| 14 | - <table id="foo"> for making columns sortable
|
| 15 | - for choosing the comparator to use!
|
| 16 | - for highlighting on sort
|
| 17 | - static / visual
|
| 18 | - Aligning right for number, left for strings.
|
| 19 | - highlighting NA numbers in red (only if it's considered a number)
|
| 20 | - formatting numbers to a certain precision
|
| 21 | - or displaying them as percentages
|
| 22 | - changing CSV headers like 'elapsed_ms' to 'elapsed ms'
|
| 23 | - Accepting a column with a '_HREF' suffix to make an HTML link
|
| 24 | - We could have something like type:
|
| 25 | string/anchor:shell-id
|
| 26 | string/href:shell-id
|
| 27 | - But the simple _HREF suffix is simpler. Easier to write R code for.
|
| 28 |
|
| 29 | Implementation notes:
|
| 30 | - To align right: need a class on every cell, e.g. "num". Can't do it through
|
| 31 | <colgroup>.
|
| 32 | - To color, can use <colgroup>. table-sort.js needs this.
|
| 33 |
|
| 34 | TODO:
|
| 35 | Does it make sense to implement <rowspan> and <colspan> ? It's nice for
|
| 36 | visualization.
|
| 37 | """
|
| 38 |
|
| 39 | import cgi
|
| 40 | import csv
|
| 41 | import optparse
|
| 42 | import os
|
| 43 | import re
|
| 44 | import sys
|
| 45 |
|
| 46 |
|
| 47 | def log(msg, *args):
|
| 48 | if args:
|
| 49 | msg = msg % args
|
| 50 | print(msg, file=sys.stderr)
|
| 51 |
|
| 52 |
|
| 53 | class NullSchema:
|
| 54 | def VerifyColumnNames(self, col_names):
|
| 55 | pass
|
| 56 |
|
| 57 | def IsNumeric(self, col_name):
|
| 58 | return False
|
| 59 |
|
| 60 | def ColumnIndexIsNumeric(self, index):
|
| 61 | return False
|
| 62 |
|
| 63 | def ColumnIndexIsInteger(self, index):
|
| 64 | return False
|
| 65 |
|
| 66 | def ColumnIndexHasHref(self, index):
|
| 67 | return False
|
| 68 |
|
| 69 |
|
| 70 | INTEGER_TYPES = ('integer',)
|
| 71 |
|
| 72 | # for sorting, right-justification
|
| 73 | NUMERIC_TYPES = ('double', 'number') + INTEGER_TYPES
|
| 74 |
|
| 75 |
|
| 76 | class Schema:
|
| 77 | def __init__(self, rows):
|
| 78 | schema_col_names = rows[0]
|
| 79 | assert 'column_name' in schema_col_names, schema_col_names
|
| 80 | assert 'type' in schema_col_names, schema_col_names
|
| 81 |
|
| 82 | # Schema columns
|
| 83 | s_cols = {}
|
| 84 | s_cols['column_name'] = []
|
| 85 | s_cols['type'] = []
|
| 86 | s_cols['precision'] = []
|
| 87 | for row in rows[1:]:
|
| 88 | for i, cell in enumerate(row):
|
| 89 | name = schema_col_names[i]
|
| 90 | s_cols[name].append(cell)
|
| 91 |
|
| 92 | self.type_lookup = dict(
|
| 93 | (name, t) for (name, t) in
|
| 94 | zip(s_cols['column_name'], s_cols['type']))
|
| 95 |
|
| 96 | # NOTE: it's OK if precision is missing.
|
| 97 | self.precision_lookup = dict(
|
| 98 | (name, p) for (name, p) in
|
| 99 | zip(s_cols['column_name'], s_cols['precision']))
|
| 100 |
|
| 101 | #log('SCHEMA %s', schema_col_names)
|
| 102 | #log('type_lookup %s', self.type_lookup)
|
| 103 | #log('precision_lookup %s', self.precision_lookup)
|
| 104 |
|
| 105 | self.col_names = None
|
| 106 | self.col_has_href = None
|
| 107 |
|
| 108 | def VerifyColumnNames(self, col_names):
|
| 109 | """Assert that the column names we got are all in the schema."""
|
| 110 | for name in col_names:
|
| 111 | log('%s : %s', name, self.type_lookup[name])
|
| 112 |
|
| 113 | n = len(col_names)
|
| 114 | self.col_has_href = [False] * n
|
| 115 | for i in xrange(n-1):
|
| 116 | this_name, next_name= col_names[i], col_names[i+1]
|
| 117 | if this_name + '_HREF' == next_name:
|
| 118 | self.col_has_href[i] = True
|
| 119 |
|
| 120 | log('href: %s', self.col_has_href)
|
| 121 | self.col_names = col_names
|
| 122 |
|
| 123 | def IsNumeric(self, col_name):
|
| 124 | return self.type_lookup[col_name] in NUMERIC_TYPES
|
| 125 |
|
| 126 | def ColumnIndexIsNumeric(self, index):
|
| 127 | col_name = self.col_names[index]
|
| 128 | return self.IsNumeric(col_name)
|
| 129 |
|
| 130 | def ColumnIndexIsInteger(self, index):
|
| 131 | col_name = self.col_names[index]
|
| 132 | return self.type_lookup[col_name] in INTEGER_TYPES
|
| 133 |
|
| 134 | def ColumnIndexHasHref(self, index):
|
| 135 | """
|
| 136 | Is the next one?
|
| 137 | """
|
| 138 | return self.col_has_href[index]
|
| 139 |
|
| 140 | def ColumnPrecision(self, index):
|
| 141 | col_name = self.col_names[index]
|
| 142 | return self.precision_lookup.get(col_name, 1) # default is arbitrary
|
| 143 |
|
| 144 |
|
| 145 | def PrintRow(row, schema):
|
| 146 | """Print a CSV row as HTML, using the given formatting.
|
| 147 |
|
| 148 | Returns:
|
| 149 | An array of booleans indicating whether each cell is a number.
|
| 150 | """
|
| 151 | i = 0
|
| 152 | n = len(row)
|
| 153 | while True:
|
| 154 | if i == n:
|
| 155 | break
|
| 156 |
|
| 157 | cell = row[i]
|
| 158 | css_classes = []
|
| 159 | cell_str = cell # by default, we don't touch it
|
| 160 |
|
| 161 | if schema.ColumnIndexIsInteger(i):
|
| 162 | css_classes.append('num') # right justify
|
| 163 |
|
| 164 | try:
|
| 165 | cell_int = int(cell)
|
| 166 | except ValueError:
|
| 167 | pass # NA?
|
| 168 | else:
|
| 169 | # commas AND floating point
|
| 170 | cell_str = '{:,}'.format(cell_int)
|
| 171 |
|
| 172 | # Look up by index now?
|
| 173 | elif schema.ColumnIndexIsNumeric(i):
|
| 174 | css_classes.append('num') # right justify
|
| 175 |
|
| 176 | try:
|
| 177 | cell_float = float(cell)
|
| 178 | except ValueError:
|
| 179 | pass # NA
|
| 180 | else:
|
| 181 | # commas AND floating point to a given precision
|
| 182 | precision = schema.ColumnPrecision(i)
|
| 183 | cell_str = '{0:,.{precision}f}'.format(cell_float, precision=precision)
|
| 184 |
|
| 185 | # Percentage
|
| 186 | #cell_str = '{:.1f}%'.format(cell_float * 100)
|
| 187 |
|
| 188 | # Special CSS class for R NA values.
|
| 189 | if cell.strip() == 'NA':
|
| 190 | css_classes.append('na') # make it red
|
| 191 |
|
| 192 | if css_classes:
|
| 193 | print(' <td class="{}">'.format(' '.join(css_classes)), end=' ')
|
| 194 | else:
|
| 195 | print(' <td>', end=' ')
|
| 196 |
|
| 197 | # Advance to next row if it's an _HREF.
|
| 198 | if schema.ColumnIndexHasHref(i):
|
| 199 | i += 1
|
| 200 | href = row[i]
|
| 201 | s = '<a href="%s">%s</a>' % (cgi.escape(href), cgi.escape(cell_str))
|
| 202 | else:
|
| 203 | s = cgi.escape(cell_str)
|
| 204 |
|
| 205 | print(s, end=' ')
|
| 206 | print('</td>')
|
| 207 |
|
| 208 | i += 1
|
| 209 |
|
| 210 |
|
| 211 | def PrintColGroup(col_names, schema):
|
| 212 | """Print HTML colgroup element, used for JavaScript sorting."""
|
| 213 | print(' <colgroup>')
|
| 214 | for i, col in enumerate(col_names):
|
| 215 | if col.endswith('_HREF'):
|
| 216 | continue
|
| 217 |
|
| 218 | # CSS class is used for sorting
|
| 219 | if schema.IsNumeric(col):
|
| 220 | css_class = 'number'
|
| 221 | else:
|
| 222 | css_class = 'case-insensitive'
|
| 223 |
|
| 224 | # NOTE: id is a comment only; not used
|
| 225 | print(' <col id="{}" type="{}" />'.format(col, css_class))
|
| 226 | print(' </colgroup>')
|
| 227 |
|
| 228 |
|
| 229 | def PrintTable(css_id, schema, col_names, rows, css_class_pattern):
|
| 230 | if css_class_pattern:
|
| 231 | css_class, r = css_class_pattern.split(None, 2)
|
| 232 | cell_regex = re.compile(r)
|
| 233 | else:
|
| 234 | css_class = None
|
| 235 | cell_regex = None
|
| 236 |
|
| 237 | print('<table id="%s">' % css_id)
|
| 238 | print(' <thead>')
|
| 239 | print(' <tr>')
|
| 240 | for i, col in enumerate(col_names):
|
| 241 | if col.endswith('_HREF'):
|
| 242 | continue
|
| 243 | heading_str = cgi.escape(col.replace('_', ' '))
|
| 244 | if schema.ColumnIndexIsNumeric(i):
|
| 245 | print(' <td class="num">%s</td>' % heading_str)
|
| 246 | else:
|
| 247 | print(' <td>%s</td>' % heading_str)
|
| 248 | print(' </tr>')
|
| 249 | print(' </thead>')
|
| 250 |
|
| 251 | print(' <tbody>')
|
| 252 | for row in rows:
|
| 253 |
|
| 254 | # TODO: There should be a special column called CSS_CLASS. Output that
|
| 255 | # from R.
|
| 256 | row_class = ''
|
| 257 | if cell_regex:
|
| 258 | for cell in row:
|
| 259 | if cell_regex.search(cell):
|
| 260 | row_class = 'class="%s"' % css_class
|
| 261 | break
|
| 262 |
|
| 263 | print(' <tr {}>'.format(row_class))
|
| 264 |
|
| 265 | PrintRow(row, schema)
|
| 266 | print(' </tr>')
|
| 267 | print(' </tbody>')
|
| 268 |
|
| 269 | PrintColGroup(col_names, schema)
|
| 270 |
|
| 271 | print('</table>')
|
| 272 |
|
| 273 |
|
| 274 | def ReadFile(f, tsv=False):
|
| 275 | """Read the CSV file, returning the column names and rows."""
|
| 276 |
|
| 277 | if tsv:
|
| 278 | c = csv.reader(f, delimiter='\t', doublequote=False,
|
| 279 | quoting=csv.QUOTE_NONE)
|
| 280 | else:
|
| 281 | c = csv.reader(f)
|
| 282 |
|
| 283 | # The first row of the CSV is assumed to be a header. The rest are data.
|
| 284 | col_names = []
|
| 285 | rows = []
|
| 286 | for i, row in enumerate(c):
|
| 287 | if i == 0:
|
| 288 | col_names = row
|
| 289 | continue
|
| 290 | rows.append(row)
|
| 291 | return col_names, rows
|
| 292 |
|
| 293 |
|
| 294 | def CreateOptionsParser():
|
| 295 | p = optparse.OptionParser()
|
| 296 |
|
| 297 | # We are taking a path, and not using stdin, because we read it twice.
|
| 298 | p.add_option(
|
| 299 | '--schema', dest='schema', metavar="PATH", type='str',
|
| 300 | help='Path to the schema.')
|
| 301 | p.add_option(
|
| 302 | '--tsv', dest='tsv', default=False, action='store_true',
|
| 303 | help='Read input in TSV format')
|
| 304 | p.add_option(
|
| 305 | '--css-class-pattern', dest='css_class_pattern', type='str',
|
| 306 | help='A string of the form CSS_CLASS:PATTERN. If the cell contents '
|
| 307 | 'matches the pattern, then apply the given CSS class. '
|
| 308 | 'Example: osh:^osh')
|
| 309 | return p
|
| 310 |
|
| 311 |
|
| 312 | def main(argv):
|
| 313 | (opts, argv) = CreateOptionsParser().parse_args(argv[1:])
|
| 314 |
|
| 315 | try:
|
| 316 | csv_path = argv[0]
|
| 317 | except IndexError:
|
| 318 | raise RuntimeError('Expected CSV filename.')
|
| 319 |
|
| 320 | schema = None
|
| 321 | if opts.schema:
|
| 322 | try:
|
| 323 | schema_f = open(opts.schema)
|
| 324 | except IOError as e:
|
| 325 | raise RuntimeError('Error opening schema: %s' % e)
|
| 326 | else:
|
| 327 | if csv_path.endswith('.csv'):
|
| 328 | schema_path = csv_path.replace('.csv', '.schema.csv')
|
| 329 | elif csv_path.endswith('.tsv'):
|
| 330 | schema_path = csv_path.replace('.tsv', '.schema.tsv')
|
| 331 | else:
|
| 332 | raise AssertionError(csv_path)
|
| 333 |
|
| 334 | log('schema path %s', schema_path)
|
| 335 | try:
|
| 336 | schema_f = open(schema_path)
|
| 337 | except IOError:
|
| 338 | schema_f = None # allowed to have no schema
|
| 339 |
|
| 340 | if schema_f:
|
| 341 | if opts.tsv:
|
| 342 | r = csv.reader(schema_f, delimiter='\t', doublequote=False,
|
| 343 | quoting=csv.QUOTE_NONE)
|
| 344 | else:
|
| 345 | r = csv.reader(schema_f)
|
| 346 |
|
| 347 | schema = Schema(list(r))
|
| 348 | else:
|
| 349 | schema = NullSchema()
|
| 350 | # Default string schema
|
| 351 |
|
| 352 | log('schema %s', schema)
|
| 353 |
|
| 354 | with open(csv_path) as f:
|
| 355 | col_names, rows = ReadFile(f, opts.tsv)
|
| 356 |
|
| 357 | schema.VerifyColumnNames(col_names)
|
| 358 |
|
| 359 | filename = os.path.basename(csv_path)
|
| 360 | css_id, _ = os.path.splitext(filename)
|
| 361 | PrintTable(css_id, schema, col_names, rows, opts.css_class_pattern)
|
| 362 |
|
| 363 |
|
| 364 | if __name__ == '__main__':
|
| 365 | try:
|
| 366 | main(sys.argv)
|
| 367 | except RuntimeError as e:
|
| 368 | print('FATAL: %s' % e, file=sys.stderr)
|
| 369 | sys.exit(1)
|