jyapayne/Nim
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

parsecsv fix style (#17933)

Browse source 
* ReSync with Devel
* Fix style in parsecsv
* Update lib/pure/parsecsv.nim

Co-authored-by: Andreas Rumpf <rumpf_a@web.de>
This commit is contained in:
Juan Carlos 2021-05-04 08:13:23 -03:00 • committed by GitHub
commit bf5dfe481a
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23
1 changed files with 64 additions and 67 deletions
Download patch file Download diff file Expand all files Collapse all files

129
lib/pure/parsecsv.nim
View file

@ -65,8 +65,7 @@
## * `parsesql module <parsesql.html>`_ for a SQL parser ## * `parsesql module <parsesql.html>`_ for a SQL parser
## * `other parsers <lib.html#pure-libraries-parsers>`_ for other parsers ## * `other parsers <lib.html#pure-libraries-parsers>`_ for other parsers
import import lexbase, streams
lexbase, streams
type type
CsvRow* = seq[string] ## A row in a CSV file. CsvRow* = seq[string] ## A row in a CSV file.
@ -97,10 +96,10 @@ proc raiseEInvalidCsv(filename: string, line, col: int,
e.msg = filename & "(" & $line & ", " & $col & ") Error: " & msg e.msg = filename & "(" & $line & ", " & $col & ") Error: " & msg
raise e raise e
proc error(my: CsvParser, pos: int, msg: string) = proc error(self: CsvParser, pos: int, msg: string) =
raiseEInvalidCsv(my.filename, my.lineNumber, getColNumber(my, pos), msg) raiseEInvalidCsv(self.filename, self.lineNumber, getColNumber(self, pos), msg)
proc open*(my: var CsvParser, input: Stream, filename: string, proc open*(self: var CsvParser, input: Stream, filename: string,
separator = ',', quote = '"', escape = '\0', separator = ',', quote = '"', escape = '\0',
skipInitialSpace = false) = skipInitialSpace = false) =
## Initializes the parser with an input stream. `Filename` is only used ## Initializes the parser with an input stream. `Filename` is only used
@ -127,16 +126,14 @@ proc open*(my: var CsvParser, input: Stream, filename: string,
parser.close() parser.close()
strm.close() strm.close()
lexbase.open(my, input) lexbase.open(self, input)
my.filename = filename self.filename = filename
my.sep = separator self.sep = separator
my.quote = quote self.quote = quote
my.esc = escape self.esc = escape
my.skipWhite = skipInitialSpace self.skipWhite = skipInitialSpace
my.row = @[]
my.currRow = 0
proc open*(my: var CsvParser, filename: string, proc open*(self: var CsvParser, filename: string,
separator = ',', quote = '"', escape = '\0', separator = ',', quote = '"', escape = '\0',
skipInitialSpace = false) = skipInitialSpace = false) =
## Similar to the `other open proc<#open,CsvParser,Stream,string,char,char,char>`_, ## Similar to the `other open proc<#open,CsvParser,Stream,string,char,char,char>`_,
@ -150,54 +147,54 @@ proc open*(my: var CsvParser, filename: string,
removeFile("tmp.csv") removeFile("tmp.csv")
var s = newFileStream(filename, fmRead) var s = newFileStream(filename, fmRead)
if s == nil: my.error(0, "cannot open: " & filename) if s == nil: self.error(0, "cannot open: " & filename)
open(my, s, filename, separator, open(self, s, filename, separator,
quote, escape, skipInitialSpace) quote, escape, skipInitialSpace)
proc parseField(my: var CsvParser, a: var string) = proc parseField(self: var CsvParser, a: var string) =
var pos = my.bufpos var pos = self.bufpos
if my.skipWhite: if self.skipWhite:
while my.buf[pos] in {' ', '\t'}: inc(pos) while self.buf[pos] in {' ', '\t'}: inc(pos)
setLen(a, 0) # reuse memory setLen(a, 0) # reuse memory
if my.buf[pos] == my.quote and my.quote != '\0': if self.buf[pos] == self.quote and self.quote != '\0':
inc(pos) inc(pos)
while true: while true:
let c = my.buf[pos] let c = self.buf[pos]
if c == '\0': if c == '\0':
my.bufpos = pos # can continue after exception? self.bufpos = pos # can continue after exception?
error(my, pos, my.quote & " expected") error(self, pos, self.quote & " expected")
break break
elif c == my.quote: elif c == self.quote:
if my.esc == '\0' and my.buf[pos+1] == my.quote: if self.esc == '\0' and self.buf[pos + 1] == self.quote:
add(a, my.quote) add(a, self.quote)
inc(pos, 2) inc(pos, 2)
else: else:
inc(pos) inc(pos)
break break
elif c == my.esc: elif c == self.esc:
add(a, my.buf[pos+1]) add(a, self.buf[pos + 1])
inc(pos, 2) inc(pos, 2)
else: else:
case c case c
of '\c': of '\c':
pos = handleCR(my, pos) pos = handleCR(self, pos)
add(a, "\n") add(a, "\n")
of '\l': of '\l':
pos = handleLF(my, pos) pos = handleLF(self, pos)
add(a, "\n") add(a, "\n")
else: else:
add(a, c) add(a, c)
inc(pos) inc(pos)
else: else:
while true: while true:
let c = my.buf[pos] let c = self.buf[pos]
if c == my.sep: break if c == self.sep: break
if c in {'\c', '\l', '\0'}: break if c in {'\c', '\l', '\0'}: break
add(a, c) add(a, c)
inc(pos) inc(pos)
my.bufpos = pos self.bufpos = pos
proc processedRows*(my: var CsvParser): int = proc processedRows*(self: var CsvParser): int {.inline.} =
## Returns number of the processed rows. ## Returns number of the processed rows.
## ##
## But even if `readRow <#readRow,CsvParser,int>`_ arrived at EOF then ## But even if `readRow <#readRow,CsvParser,int>`_ arrived at EOF then
@ -220,9 +217,9 @@ proc processedRows*(my: var CsvParser): int =
parser.close() parser.close()
strm.close() strm.close()
return my.currRow self.currRow
proc readRow*(my: var CsvParser, columns = 0): bool = proc readRow*(self: var CsvParser, columns = 0): bool =
## Reads the next row; if `columns` > 0, it expects the row to have ## Reads the next row; if `columns` > 0, it expects the row to have
## exactly this many columns. Returns false if the end of the file ## exactly this many columns. Returns false if the end of the file
## has been encountered else true. ## has been encountered else true.
@ -251,47 +248,47 @@ proc readRow*(my: var CsvParser, columns = 0): bool =
strm.close() strm.close()
var col = 0 # current column var col = 0 # current column
let oldpos = my.bufpos let oldpos = self.bufpos
# skip initial empty lines #8365 # skip initial empty lines #8365
while true: while true:
case my.buf[my.bufpos] case self.buf[self.bufpos]
of '\c': my.bufpos = handleCR(my, my.bufpos) of '\c': self.bufpos = handleCR(self, self.bufpos)
of '\l': my.bufpos = handleLF(my, my.bufpos) of '\l': self.bufpos = handleLF(self, self.bufpos)
else: break else: break
while my.buf[my.bufpos] != '\0': while self.buf[self.bufpos] != '\0':
let oldlen = my.row.len let oldlen = self.row.len
if oldlen < col+1: if oldlen < col + 1:
setLen(my.row, col+1) setLen(self.row, col + 1)
my.row[col] = "" self.row[col] = ""
parseField(my, my.row[col]) parseField(self, self.row[col])
inc(col) inc(col)
if my.buf[my.bufpos] == my.sep: if self.buf[self.bufpos] == self.sep:
inc(my.bufpos) inc(self.bufpos)
else: else:
case my.buf[my.bufpos] case self.buf[self.bufpos]
of '\c', '\l': of '\c', '\l':
# skip empty lines: # skip empty lines:
while true: while true:
case my.buf[my.bufpos] case self.buf[self.bufpos]
of '\c': my.bufpos = handleCR(my, my.bufpos) of '\c': self.bufpos = handleCR(self, self.bufpos)
of '\l': my.bufpos = handleLF(my, my.bufpos) of '\l': self.bufpos = handleLF(self, self.bufpos)
else: break else: break
of '\0': discard of '\0': discard
else: error(my, my.bufpos, my.sep & " expected") else: error(self, self.bufpos, self.sep & " expected")
break break
setLen(my.row, col) setLen(self.row, col)
result = col > 0 result = col > 0
if result and col != columns and columns > 0: if result and col != columns and columns > 0:
error(my, oldpos+1, $columns & " columns expected, but found " & error(self, oldpos + 1, $columns & " columns expected, but found " &
$col & " columns") $col & " columns")
inc(my.currRow) inc(self.currRow)
proc close*(my: var CsvParser) {.inline.} = proc close*(self: var CsvParser) {.inline.} =
## Closes the parser `my` and its associated input stream. ## Closes the parser `self` and its associated input stream.
lexbase.close(my) lexbase.close(self)
proc readHeaderRow*(my: var CsvParser) = proc readHeaderRow*(self: var CsvParser) =
## Reads the first row and creates a look-up table for column numbers ## Reads the first row and creates a look-up table for column numbers
## See also: ## See also:
## * `rowEntry proc <#rowEntry,CsvParser,string>`_ ## * `rowEntry proc <#rowEntry,CsvParser,string>`_
@ -313,11 +310,11 @@ proc readHeaderRow*(my: var CsvParser) =
parser.close() parser.close()
strm.close() strm.close()
let present = my.readRow() let present = self.readRow()
if present: if present:
my.headers = my.row self.headers = self.row
proc rowEntry*(my: var CsvParser, entry: string): var string = proc rowEntry*(self: var CsvParser, entry: string): var string =
## Accesses a specified `entry` from the current row. ## Accesses a specified `entry` from the current row.
## ##
## Assumes that `readHeaderRow <#readHeaderRow,CsvParser>`_ has already been ## Assumes that `readHeaderRow <#readHeaderRow,CsvParser>`_ has already been
@ -340,9 +337,9 @@ proc rowEntry*(my: var CsvParser, entry: string): var string =
parser.close() parser.close()
strm.close() strm.close()
let index = my.headers.find(entry) let index = self.headers.find(entry)
if index >= 0: if index >= 0:
result = my.row[index] result = self.row[index]
else: else:
raise newException(KeyError, "Entry `" & entry & "` doesn't exist") raise newException(KeyError, "Entry `" & entry & "` doesn't exist")