cpp/ql/lib/semmle/code/cpp/AutogeneratedFile.qll

/**
 * Provides a class and predicate for recognizing files that are likely to have been generated
 * automatically.
 */

import semmle.code.cpp.Comments
import semmle.code.cpp.File
import semmle.code.cpp.Preprocessor

/**
 * Holds if comment `c` indicates that it might be in an auto-generated file, for
 * example because it contains the text "auto-generated by".
 */
bindingset[comment]
private predicate autogeneratedComment(string comment) {
  // ?s = include newlines in anything (`.`)
  // ?i = ignore case
  exists(string cond |
    cond =
      // generated by (not mid-sentence)
      "(^ generated by[^a-z])|" + "(! generated by[^a-z])|" +
        // generated file
        "(generated file)|" +
        // file [is/was/has been] generated
        "(file( is| was| has been)? generated)|" +
        // changes made in this file will be lost
        "(changes made in this file will be lost)|" +
        // do not edit/modify (not mid-sentence)
        "(^ do(n't|nt| not) (hand-?)?(edit|modify))|" +
        "(! do(n't|nt| not) (hand-?)?(edit|modify))|" +
        // do not edit/modify + generated
        "(do(n't|nt| not) (hand-?)?(edit|modify).*generated)|" +
        "(generated.*do(n't|nt| not) (hand-?)?(edit|modify))" and
    comment
        .regexpMatch("(?si).*(" +
            // replace `generated` with a regexp that also catches things like
            // `auto-generated`.
            cond.replaceAll("generated", "(auto[\\w-]*[\\s/\\*\\r\\n]*)?generated")
                // replace `!` with a regexp for end-of-sentence / separator characters.
                .replaceAll("!", "[\\.\\?\\!\\-\\;\\,]")
                // replace ` ` with a regexp for one or more whitespace characters
                // (including newlines and `/*`).
                .replaceAll(" ", "[\\s/\\*\\r\\n]+") + ").*")
  )
}

/**
 * Holds if the file contains `#line` pragmas that refer to a different file.
 * For example, in `parser.c` a pragma `#line 1 "parser.rl"`.
 * Such pragmas usually indicate that the file was automatically generated.
 */
predicate hasPragmaDifferentFile(File f) {
  exists(PreprocessorLine pl, string s |
    pl.getFile() = f and
    pl.getHead().splitAt(" ", 1) = s and
    /* Zero index is line number, one index is file reference */
    not "\"" + f.getAbsolutePath() + "\"" = s and
    not "\"" + f.getRelativePath() + "\"" = s and
    not "\"" + f.getBaseName() + "\"" = s
  )
}

/**
 * The line where the first comment in file `f` begins (maximum of 5).  This allows
 * us to skip past any preprocessor logic or similar code before the first comment.
 */
private int fileFirstComment(File f) {
  result =
    min(int line |
      exists(Comment c |
        c.getFile() = f and
        c.getLocation().getStartLine() = line and
        line < 5
      )
    ).minimum(5)
}

/**
 * The line where the initial comments of file `f` end.  This is just before the
 * first bit of code, excluding anything skipped over by `fileFirstComment`.
 */
private int fileHeaderLimit(File f) {
  exists(int fc |
    fc = fileFirstComment(f) and
    result =
      min(int line |
        // code ending the initial comments
        exists(DeclarationEntry de, Location l |
          l = de.getLocation() and
          l.getFile() = f and
          line = l.getStartLine() - 1 and
          line > fc
        )
        or
        exists(PreprocessorDirective pd, Location l |
          l = pd.getLocation() and
          l.getFile() = f and
          line = l.getStartLine() - 1 and
          line > fc
        )
        or
        exists(NamespaceDeclarationEntry nde, Location l |
          l = nde.getLocation() and
          l.getFile() = f and
          line = l.getStartLine() - 1 and
          line > fc
        )
        or
        // end of the file
        line = f.getMetrics().getNumberOfLines()
        or
        // rarely, we've seen extremely long sequences of initial comments
        // (and/or limitations in the above constraints) cause an overflow of
        // the maximum string length. So don't look past 1000 lines regardless.
        line = 1000
      )
  )
}

/**
 * Holds if the file is probably an autogenerated file.
 *
 * A file is probably autogenerated if either of the following heuristics
 * hold:
 *   1. There is a comment in the start of the file that matches
 *      'autogenerated', 'generated by', or a similar phrase.
 *   2. There is a `#line` directive referring to a different file.
 */
class AutogeneratedFile extends File {
  cached
  AutogeneratedFile() {
    autogeneratedComment(strictconcat(Comment c |
        c.getFile() = this and
        c.getLocation().getStartLine() <= fileHeaderLimit(this)
      |
        c.getContents() order by c.getLocation().getStartLine()
      )) or
    hasPragmaDifferentFile(this)
  }
}