Skip to main content

PCRE – Perl Compatible Regular Expressions

Overview​

PCRE is a full-featured regular expression library that implements the Perl syntax and semantics. It provides powerful pattern matching capabilities, including lookarounds, backreferences, and advanced capture handling.

PCRE uses a backtracking engine, which allows flexible and expressive regex patterns but can lead to exponential execution time for certain pathological patterns. It is best suited for controlled environments where input patterns are trusted, such as batch processing or offline text analysis.

Key Features​

  • Full Perl syntax support: Lookaheads, lookbehinds, backreferences, and conditional patterns.
  • Dynamic compilation: Regex patterns can be compiled at runtime.
  • Capture groups: Supports numbered and named captures.
  • Unicode support: Handles UTF-8, UTF-16, and other encodings.
  • Flexible matching options: Partial matches, global matches, and custom flags.

Typical Use Cases​

  • Offline text processing and transformation.
  • Advanced pattern matching for logs, documents, or corpora.
  • Controlled batch analytics where performance guarantees are not critical.

Industrial Fit​

RequirementPCRE Support
Full Perl regex syntax✔️
Dynamic pattern compilation✔️
Safe for untrusted input❌
Predictable performance❌
Capture groups✔️
Unicode support✔️

C++ Example​

#include <pcre.h>
#include <iostream>
#include <string>

int main() {
    const char* pattern = "User: (\\w+), Score: (\\d+)";
    const char* subject = "User: Alice, Score: 42";
    const char* error;
    int erroffset;

    pcre* re = pcre_compile(pattern, 0, &error, &erroffset, nullptr);
    if (!re) return 1;

    int ovector[30];
    int rc = pcre_exec(re, nullptr, subject, strlen(subject), 0, 0, ovector, 30);
    if (rc >= 0) {
        std::cout.write(subject + ovector[2], ovector[3] - ovector[2]);
        std::cout << " ";
        std::cout.write(subject + ovector[4], ovector[5] - ovector[4]);
        std::cout << std::endl;
    }

    pcre_free(re);
    return 0;
}