// Copyright (c) 2005, Google Inc.

[0] Fix | Delete

// All rights reserved.

[1] Fix | Delete

//

[2] Fix | Delete

// Redistribution and use in source and binary forms, with or without

[3] Fix | Delete

// modification, are permitted provided that the following conditions are

[4] Fix | Delete

// met:

[5] Fix | Delete

//

[6] Fix | Delete

// * Redistributions of source code must retain the above copyright

[7] Fix | Delete

// notice, this list of conditions and the following disclaimer.

[8] Fix | Delete

// * Redistributions in binary form must reproduce the above

[9] Fix | Delete

// copyright notice, this list of conditions and the following disclaimer

[10] Fix | Delete

// in the documentation and/or other materials provided with the

[11] Fix | Delete

// distribution.

[12] Fix | Delete

// * Neither the name of Google Inc. nor the names of its

[13] Fix | Delete

// contributors may be used to endorse or promote products derived from

[14] Fix | Delete

// this software without specific prior written permission.

[15] Fix | Delete

//

[16] Fix | Delete

// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS

[17] Fix | Delete

// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT

[18] Fix | Delete

// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR

[19] Fix | Delete

// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT

[20] Fix | Delete

// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,

[21] Fix | Delete

// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT

[22] Fix | Delete

// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,

[23] Fix | Delete

// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY

[24] Fix | Delete

// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT

[25] Fix | Delete

// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE

[26] Fix | Delete

// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.

[27] Fix | Delete

//

[28] Fix | Delete

// Author: Sanjay Ghemawat

[29] Fix | Delete

//

[30] Fix | Delete

// Regular-expression based scanner for parsing an input stream.

[31] Fix | Delete

//

[32] Fix | Delete

// Example 1: parse a sequence of "var = number" entries from input:

[33] Fix | Delete

//

[34] Fix | Delete

// Scanner scanner(input);

[35] Fix | Delete

// string var;

[36] Fix | Delete

// int number;

[37] Fix | Delete

// scanner.SetSkipExpression("\\s+"); // Skip any white space we encounter

[38] Fix | Delete

// while (scanner.Consume("(\\w+) = (\\d+)", &var, &number)) {

[39] Fix | Delete

// ...;

[40] Fix | Delete

// }

[41] Fix | Delete

[42] Fix | Delete

#ifndef _PCRE_SCANNER_H

[43] Fix | Delete

#define _PCRE_SCANNER_H

[44] Fix | Delete

[45] Fix | Delete

#include <assert.h>

[46] Fix | Delete

#include <string>

[47] Fix | Delete

#include <vector>

[48] Fix | Delete

[49] Fix | Delete

#include <pcrecpp.h>

[50] Fix | Delete

#include <pcre_stringpiece.h>

[51] Fix | Delete

[52] Fix | Delete

namespace pcrecpp {

[53] Fix | Delete

[54] Fix | Delete

class PCRECPP_EXP_DEFN Scanner {

[55] Fix | Delete

public:

[56] Fix | Delete

Scanner();

[57] Fix | Delete

explicit Scanner(const std::string& input);

[58] Fix | Delete

~Scanner();

[59] Fix | Delete

[60] Fix | Delete

// Return current line number. The returned line-number is

[61] Fix | Delete

// one-based. I.e. it returns 1 + the number of consumed newlines.

[62] Fix | Delete

//

[63] Fix | Delete

// Note: this method may be slow. It may take time proportional to

[64] Fix | Delete

// the size of the input.

[65] Fix | Delete

int LineNumber() const;

[66] Fix | Delete

[67] Fix | Delete

// Return the byte-offset that the scanner is looking in the

[68] Fix | Delete

// input data;

[69] Fix | Delete

int Offset() const;

[70] Fix | Delete

[71] Fix | Delete

// Return true iff the start of the remaining input matches "re"

[72] Fix | Delete

bool LookingAt(const RE& re) const;

[73] Fix | Delete

[74] Fix | Delete

// Return true iff all of the following are true

[75] Fix | Delete

// a. the start of the remaining input matches "re",

[76] Fix | Delete

// b. if any arguments are supplied, matched sub-patterns can be

[77] Fix | Delete

// parsed and stored into the arguments.

[78] Fix | Delete

// If it returns true, it skips over the matched input and any

[79] Fix | Delete

// following input that matches the "skip" regular expression.

[80] Fix | Delete

bool Consume(const RE& re,

[81] Fix | Delete

const Arg& arg0 = RE::no_arg,

[82] Fix | Delete

const Arg& arg1 = RE::no_arg,

[83] Fix | Delete

const Arg& arg2 = RE::no_arg

[84] Fix | Delete

// TODO: Allow more arguments?

[85] Fix | Delete

);

[86] Fix | Delete

[87] Fix | Delete

// Set the "skip" regular expression. If after consuming some data,

[88] Fix | Delete

// a prefix of the input matches this RE, it is automatically

[89] Fix | Delete

// skipped. For example, a programming language scanner would use

[90] Fix | Delete

// a skip RE that matches white space and comments.

[91] Fix | Delete

//

[92] Fix | Delete

// scanner.SetSkipExpression("\\s+|//.*|/[*](.|\n)*?[*]/");

[93] Fix | Delete

//

[94] Fix | Delete

// Skipping repeats as long as it succeeds. We used to let people do

[95] Fix | Delete

// this by writing "(...)*" in the regular expression, but that added

[96] Fix | Delete

// up to lots of recursive calls within the pcre library, so now we

[97] Fix | Delete

// control repetition explicitly via the function call API.

[98] Fix | Delete

//

[99] Fix | Delete

// You can pass NULL for "re" if you do not want any data to be skipped.

[100] Fix | Delete

void Skip(const char* re); // DEPRECATED; does *not* repeat

[101] Fix | Delete

void SetSkipExpression(const char* re);

[102] Fix | Delete

[103] Fix | Delete

// Temporarily pause "skip"ing. This

[104] Fix | Delete

// Skip("Foo"); code ; DisableSkip(); code; EnableSkip()

[105] Fix | Delete

// is similar to

[106] Fix | Delete

// Skip("Foo"); code ; Skip(NULL); code ; Skip("Foo");

[107] Fix | Delete

// but avoids creating/deleting new RE objects.

[108] Fix | Delete

void DisableSkip();

[109] Fix | Delete

[110] Fix | Delete

// Reenable previously paused skipping. Any prefix of the input

[111] Fix | Delete

// that matches the skip pattern is immediately dropped.

[112] Fix | Delete

void EnableSkip();

[113] Fix | Delete

[114] Fix | Delete

/***** Special wrappers around SetSkip() for some common idioms *****/

[115] Fix | Delete

[116] Fix | Delete

// Arranges to skip whitespace, C comments, C++ comments.

[117] Fix | Delete

// The overall RE is a disjunction of the following REs:

[118] Fix | Delete

// \\s whitespace

[119] Fix | Delete

// //.*\n C++ comment

[120] Fix | Delete

// /[*](.|\n)*?[*]/ C comment (x*? means minimal repetitions of x)

[121] Fix | Delete

// We get repetition via the semantics of SetSkipExpression, not by using *

[122] Fix | Delete

void SkipCXXComments() {

[123] Fix | Delete

SetSkipExpression("\\s|//.*\n|/[*](?:\n|.)*?[*]/");

[124] Fix | Delete

}

[125] Fix | Delete

[126] Fix | Delete

void set_save_comments(bool comments) {

[127] Fix | Delete

save_comments_ = comments;

[128] Fix | Delete

}

[129] Fix | Delete

[130] Fix | Delete

bool save_comments() {

[131] Fix | Delete

return save_comments_;

[132] Fix | Delete

}

[133] Fix | Delete

[134] Fix | Delete

// Append to vector ranges the comments found in the

[135] Fix | Delete

// byte range [start,end] (inclusive) of the input data.

[136] Fix | Delete

// Only comments that were extracted entirely within that

[137] Fix | Delete

// range are returned: no range splitting of atomically-extracted

[138] Fix | Delete

// comments is performed.

[139] Fix | Delete

void GetComments(int start, int end, std::vector<StringPiece> *ranges);

[140] Fix | Delete

[141] Fix | Delete

// Append to vector ranges the comments added

[142] Fix | Delete

// since the last time this was called. This

[143] Fix | Delete

// functionality is provided for efficiency when

[144] Fix | Delete

// interleaving scanning with parsing.

[145] Fix | Delete

void GetNextComments(std::vector<StringPiece> *ranges);

[146] Fix | Delete

[147] Fix | Delete

private:

[148] Fix | Delete

std::string data_; // All the input data

[149] Fix | Delete

StringPiece input_; // Unprocessed input

[150] Fix | Delete

RE* skip_; // If non-NULL, RE for skipping input

[151] Fix | Delete

bool should_skip_; // If true, use skip_

[152] Fix | Delete

bool skip_repeat_; // If true, repeat skip_ as long as it works

[153] Fix | Delete

bool save_comments_; // If true, aggregate the skip expression

[154] Fix | Delete

[155] Fix | Delete

// the skipped comments

[156] Fix | Delete

// TODO: later consider requiring that the StringPieces be added

[157] Fix | Delete

// in order by their start position

[158] Fix | Delete

std::vector<StringPiece> *comments_;

[159] Fix | Delete

[160] Fix | Delete

// the offset into comments_ that has been returned by GetNextComments

[161] Fix | Delete

int comments_offset_;

[162] Fix | Delete

[163] Fix | Delete

// helper function to consume *skip_ and honour

[164] Fix | Delete

// save_comments_

[165] Fix | Delete

void ConsumeSkip();

[166] Fix | Delete

};

[167] Fix | Delete

[168] Fix | Delete

} // namespace pcrecpp

[169] Fix | Delete

[170] Fix | Delete

#endif /* _PCRE_SCANNER_H */

[171] Fix | Delete

[172] Fix | Delete