diff options
Diffstat (limited to 'lib/pure/lexbase.nim')
| -rw-r--r-- | lib/pure/lexbase.nim | 92 |
1 files changed, 41 insertions, 51 deletions
diff --git a/lib/pure/lexbase.nim b/lib/pure/lexbase.nim index e38acd5ef..1efd97b24 100644 --- a/lib/pure/lexbase.nim +++ b/lib/pure/lexbase.nim @@ -1,6 +1,6 @@ # # -# The Nim Compiler +# Nim's Runtime Library # (c) Copyright 2009 Andreas Rumpf # # See the file "copying.txt", included in this @@ -12,10 +12,13 @@ ## needs refilling. import - strutils, streams + std/[strutils, streams] + +when defined(nimPreviewSlimSystem): + import std/assertions const - EndOfFile* = '\0' ## end of file marker + EndOfFile* = '\0' ## end of file marker NewLines* = {'\c', '\L'} # Buffer handling: @@ -27,26 +30,17 @@ const type BaseLexer* = object of RootObj ## the base lexer. Inherit your lexer from ## this object. - bufpos*: int ## the current position within the buffer - when defined(js): ## the buffer itself - buf*: string - else: - buf*: cstring - bufLen*: int ## length of buffer in characters - input: Stream ## the input stream - lineNumber*: int ## the current line number + bufpos*: int ## the current position within the buffer + buf*: string ## the buffer itself + input: Stream ## the input stream + lineNumber*: int ## the current line number sentinel: int - lineStart: int # index of last line start in buffer - offsetBase*: int # use ``offsetBase + bufpos`` to get the offset + lineStart: int # index of last line start in buffer + offsetBase*: int # use `offsetBase + bufpos` to get the offset refillChars: set[char] -const - chrSize = sizeof(char) - proc close*(L: var BaseLexer) = ## closes the base lexer. This closes `L`'s associated stream too. - when not defined(js): - dealloc(L.buf) close(L.input) proc fillBuffer(L: var BaseLexer) = @@ -57,26 +51,31 @@ proc fillBuffer(L: var BaseLexer) = oldBufLen: int # we know here that pos == L.sentinel, but not if this proc # is called the first time by initBaseLexer() - assert(L.sentinel < L.bufLen) - toCopy = L.bufLen - L.sentinel - 1 + assert(L.sentinel + 1 <= L.buf.len) + toCopy = L.buf.len - (L.sentinel + 1) assert(toCopy >= 0) if toCopy > 0: - when defined(js): - for i in 0 ..< toCopy: L.buf[i] = L.buf[L.sentinel + 1 + i] + when defined(js) or defined(nimscript): + # nimscript has to be here to avoid compiling other branch (moveMem) + for i in 0 ..< toCopy: + L.buf[i] = L.buf[L.sentinel + 1 + i] else: - # "moveMem" handles overlapping regions - moveMem(L.buf, addr L.buf[L.sentinel + 1], toCopy * chrSize) - charsRead = readData(L.input, addr(L.buf[toCopy]), - (L.sentinel + 1) * chrSize) div chrSize + when nimvm: + for i in 0 ..< toCopy: + L.buf[i] = L.buf[L.sentinel + 1 + i] + else: + # "moveMem" handles overlapping regions + moveMem(addr L.buf[0], addr L.buf[L.sentinel + 1], toCopy) + charsRead = L.input.readDataStr(L.buf, toCopy ..< toCopy + L.sentinel + 1) s = toCopy + charsRead if charsRead < L.sentinel + 1: - L.buf[s] = EndOfFile # set end marker + L.buf[s] = EndOfFile # set end marker L.sentinel = s else: # compute sentinel: - dec(s) # BUGFIX (valgrind) + dec(s) # BUGFIX (valgrind) while true: - assert(s < L.bufLen) + assert(s < L.buf.len) while s >= 0 and L.buf[s] notin L.refillChars: dec(s) if s >= 0: # we found an appropriate character for a sentinel: @@ -85,25 +84,19 @@ proc fillBuffer(L: var BaseLexer) = else: # rather than to give up here because the line is too long, # double the buffer's size and try again: - oldBufLen = L.bufLen - L.bufLen = L.bufLen * 2 - when defined(js): - L.buf.setLen(L.bufLen) - else: - L.buf = cast[cstring](realloc(L.buf, L.bufLen * chrSize)) - assert(L.bufLen - oldBufLen == oldBufLen) - charsRead = readData(L.input, addr(L.buf[oldBufLen]), - oldBufLen * chrSize) div chrSize + oldBufLen = L.buf.len + L.buf.setLen(L.buf.len * 2) + charsRead = readDataStr(L.input, L.buf, oldBufLen ..< L.buf.len) if charsRead < oldBufLen: L.buf[oldBufLen + charsRead] = EndOfFile L.sentinel = oldBufLen + charsRead break - s = L.bufLen - 1 + s = L.buf.len - 1 proc fillBaseLexer(L: var BaseLexer, pos: int): int = assert(pos <= L.sentinel) if pos < L.sentinel: - result = pos + 1 # nothing to do + result = pos + 1 # nothing to do else: fillBuffer(L) L.offsetBase += pos @@ -111,9 +104,9 @@ proc fillBaseLexer(L: var BaseLexer, pos: int): int = result = 0 proc handleCR*(L: var BaseLexer, pos: int): int = - ## Call this if you scanned over '\c' in the buffer; it returns the the + ## Call this if you scanned over `'\c'` in the buffer; it returns the ## position to continue the scanning from. `pos` must be the position - ## of the '\c'. + ## of the `'\c'`. assert(L.buf[pos] == '\c') inc(L.lineNumber) result = fillBaseLexer(L, pos) @@ -122,16 +115,17 @@ proc handleCR*(L: var BaseLexer, pos: int): int = L.lineStart = result proc handleLF*(L: var BaseLexer, pos: int): int = - ## Call this if you scanned over '\L' in the buffer; it returns the the + ## Call this if you scanned over `'\L'` in the buffer; it returns the ## position to continue the scanning from. `pos` must be the position - ## of the '\L'. + ## of the `'\L'`. assert(L.buf[pos] == '\L') inc(L.lineNumber) result = fillBaseLexer(L, pos) #L.lastNL := result-1; // BUGFIX: was: result; L.lineStart = result proc handleRefillChar*(L: var BaseLexer, pos: int): int = - ## To be documented. + ## Call this if a terminator character other than a new line is scanned + ## at `pos`; it returns the position to continue the scanning from. assert(L.buf[pos] in L.refillChars) result = fillBaseLexer(L, pos) #L.lastNL := result-1; // BUGFIX: was: result; @@ -148,15 +142,11 @@ proc open*(L: var BaseLexer, input: Stream, bufLen: int = 8192; L.input = input L.bufpos = 0 L.offsetBase = 0 - L.bufLen = bufLen L.refillChars = refillChars - when defined(js): - L.buf = newString(bufLen) - else: - L.buf = cast[cstring](alloc(bufLen * chrSize)) + L.buf = newString(bufLen) L.sentinel = bufLen - 1 L.lineStart = 0 - L.lineNumber = 1 # lines start at 1 + L.lineNumber = 1 # lines start at 1 fillBuffer(L) skipUtf8Bom(L) |