source: icGREP/icgrep-devel/icgrep/re/re_toolchain.cpp @ 5357

Last change on this file since 5357 was 5357, checked in by xuedongx, 2 years ago

delete the linebreak computation in icgrep

File size: 4.7 KB
Line 
1/*
2 *  Copyright (c) 2015 International Characters.
3 *  This software is licensed to the public under the Open Software License 3.0.
4 *  icgrep is a trademark of International Characters.
5 */
6
7#include <re/re_toolchain.h>
8#include <cc/cc_compiler.h>            // for CC_Compiler
9#include <llvm/Support/CommandLine.h>  // for clEnumVal, clEnumValEnd, Optio...
10#include <re/re_compiler.h>            // for RE_Compiler
11#include <re/re_nullable.h>            // for RE_Nullable
12#include <re/re_simplifier.h>          // for RE_Simplifier
13#include <re/printer_re.h>
14#include <iostream>
15
16using namespace pablo;
17using namespace llvm;
18
19namespace re {
20
21static cl::OptionCategory RegexOptions("Regex Toolchain Options",
22                                              "These options control the regular expression transformation and compilation.");
23const cl::OptionCategory * re_toolchain_flags() {
24    return &RegexOptions;
25}
26
27static cl::bits<RE_PrintFlags> 
28    PrintOptions(cl::values(clEnumVal(ShowREs, "Print parsed or generated regular expressions"),
29                            clEnumVal(ShowAllREs, "Print all regular expression passes"),
30                            clEnumVal(ShowStrippedREs, "Print REs with nullable prefixes/suffixes removed"),
31                            clEnumVal(ShowSimplifiedREs, "Print final simplified REs"),
32                            clEnumValEnd), cl::cat(RegexOptions));
33
34static cl::bits<RE_AlgorithmFlags>
35    AlgorithmOptions(cl::values(clEnumVal(DisableLog2BoundedRepetition, "disable log2 optimizations for bounded repetition of bytes"),
36                              clEnumVal(DisableIfHierarchy, "disable nested if hierarchy for generated Unicode classes (not recommended)"), 
37                              clEnumVal(DisableMatchStar, "disable MatchStar optimization"), 
38                              clEnumVal(DisableUnicodeMatchStar, "disable Unicode MatchStar optimization"),
39                              clEnumVal(DisableUnicodeLineBreak, "disable Unicode line breaks - use LF only"),
40                              clEnumValN(InvertMatches, "v", "select non-matching lines"),
41                              clEnumValEnd), 
42                   cl::cat(RegexOptions));
43
44bool AlgorithmOptionIsSet(RE_AlgorithmFlags flag) {
45    return AlgorithmOptions.isSet(flag);
46}
47
48int IfInsertionGap;
49static cl::opt<int, true> 
50    IfInsertionGapOption("if-insertion-gap",  cl::location(IfInsertionGap), cl::init(3),
51                         cl::desc("minimum number of nonempty elements between inserted if short-circuit tests"), 
52                         cl::cat(RegexOptions));
53
54
55
56RE * regular_expression_passes(RE * re_ast)  {
57    if (PrintOptions.isSet(ShowAllREs) || PrintOptions.isSet(ShowREs)) {
58        std::cerr << "Parser:" << std::endl << Printer_RE::PrintRE(re_ast) << std::endl;
59    }
60
61    //Optimization passes to simplify the AST.
62    re_ast = re::RE_Nullable::removeNullablePrefix(re_ast);
63    if (PrintOptions.isSet(ShowAllREs) || PrintOptions.isSet(ShowStrippedREs)) {
64        std::cerr << "RemoveNullablePrefix:" << std::endl << Printer_RE::PrintRE(re_ast) << std::endl;
65    }
66    re_ast = re::RE_Nullable::removeNullableSuffix(re_ast);
67    if (PrintOptions.isSet(ShowAllREs) || PrintOptions.isSet(ShowStrippedREs)) {
68        std::cerr << "RemoveNullableSuffix:" << std::endl << Printer_RE::PrintRE(re_ast) << std::endl;
69    }
70    re_ast = re::RE_Nullable::removeNullableAssertion(re_ast);
71    if (PrintOptions.isSet(ShowAllREs) || PrintOptions.isSet(ShowStrippedREs)) {
72        std::cerr << "RemoveNullableAssertion:" << std::endl << Printer_RE::PrintRE(re_ast) << std::endl;
73    }
74    //re_ast = re::RE_Nullable::removeNullableAfterAssertion(re_ast);
75    //if (PrintOptions.isSet(ShowAllREs) || PrintOptions.isSet(ShowStrippedREs)) {
76    //    std::cerr << "RemoveNullableAfterAssertion" << std::endl << Printer_RE::PrintRE(re_ast) << std::endl;
77    //}
78   
79    re_ast = re::RE_Simplifier::simplify(re_ast);
80    if (PrintOptions.isSet(ShowAllREs) || PrintOptions.isSet(ShowSimplifiedREs)) {
81        //Print to the terminal the AST that was generated by the simplifier.
82        std::cerr << "Simplifier:" << std::endl << Printer_RE::PrintRE(re_ast) << std::endl;
83    }
84    return re_ast;
85}
86   
87void re2pablo_compiler(PabloKernel * kernel, RE * re_ast, const bool CountOnly) {
88    Var * const basis = kernel->getInputStreamVar("basis");
89    Var * const linebreak = kernel->getInputStreamVar("linebreak");
90    cc::CC_Compiler cc_compiler(kernel, basis);
91    re::RE_Compiler re_compiler(kernel, cc_compiler, CountOnly);
92    re_compiler.initializeRequiredStreams(basis->getType()->getArrayNumElements(), linebreak);
93    re_compiler.compileUnicodeNames(re_ast);
94    re_compiler.finalizeMatchResult(re_compiler.compile(re_ast), AlgorithmOptions.isSet(InvertMatches));
95}
96
97}
Note: See TracBrowser for help on using the repository browser.