pods.go

     1  // Copyright 2022 The Go Authors. All rights reserved.
     2  // Use of this source code is governed by a BSD-style
     3  // license that can be found in the LICENSE file.
     4  
     5  package pods
     6  
     7  import (
     8  	"fmt"
     9  	"internal/coverage"
    10  	"os"
    11  	"path/filepath"
    12  	"regexp"
    13  	"sort"
    14  	"strconv"
    15  )
    16  
    17  // Pod encapsulates a set of files emitted during the executions of a
    18  // coverage-instrumented binary. Each pod contains a single meta-data
    19  // file, and then 0 or more counter data files that refer to that
    20  // meta-data file. Pods are intended to simplify processing of
    21  // coverage output files in the case where we have several coverage
    22  // output directories containing output files derived from more
    23  // than one instrumented executable. In the case where the files that
    24  // make up a pod are spread out across multiple directories, each
    25  // element of the "Origins" field below will be populated with the
    26  // index of the originating directory for the corresponding counter
    27  // data file (within the slice of input dirs handed to CollectPods).
    28  // The ProcessIDs field will be populated with the process ID of each
    29  // data file in the CounterDataFiles slice.
    30  type Pod struct {
    31  	MetaFile         string
    32  	CounterDataFiles []string
    33  	Origins          []int
    34  	ProcessIDs       []int
    35  }
    36  
    37  // CollectPods visits the files contained within the directories in
    38  // the list 'dirs', collects any coverage-related files, partitions
    39  // them into pods, and returns a list of the pods to the caller, along
    40  // with an error if something went wrong during directory/file
    41  // reading.
    42  //
    43  // CollectPods skips over any file that is not related to coverage
    44  // (e.g. avoids looking at things that are not meta-data files or
    45  // counter-data files). CollectPods also skips over 'orphaned' counter
    46  // data files (e.g. counter data files for which we can't find the
    47  // corresponding meta-data file). If "warn" is true, CollectPods will
    48  // issue warnings to stderr when it encounters non-fatal problems (for
    49  // orphans or a directory with no meta-data files).
    50  func CollectPods(dirs []string, warn bool) ([]Pod, error) {
    51  	files := []string{}
    52  	dirIndices := []int{}
    53  	for k, dir := range dirs {
    54  		dents, err := os.ReadDir(dir)
    55  		if err != nil {
    56  			return nil, err
    57  		}
    58  		for _, e := range dents {
    59  			if e.IsDir() {
    60  				continue
    61  			}
    62  			files = append(files, filepath.Join(dir, e.Name()))
    63  			dirIndices = append(dirIndices, k)
    64  		}
    65  	}
    66  	return collectPodsImpl(files, dirIndices, warn), nil
    67  }
    68  
    69  // CollectPodsFromFiles functions the same as "CollectPods" but
    70  // operates on an explicit list of files instead of a directory.
    71  func CollectPodsFromFiles(files []string, warn bool) []Pod {
    72  	return collectPodsImpl(files, nil, warn)
    73  }
    74  
    75  type fileWithAnnotations struct {
    76  	file   string
    77  	origin int
    78  	pid    int
    79  }
    80  
    81  type protoPod struct {
    82  	mf       string
    83  	elements []fileWithAnnotations
    84  }
    85  
    86  // collectPodsImpl examines the specified list of files and picks out
    87  // subsets that correspond to coverage pods. The first stage in this
    88  // process is collecting a set { M1, M2, ... MN } where each M_k is a
    89  // distinct coverage meta-data file. We then create a single pod for
    90  // each meta-data file M_k, then find all of the counter data files
    91  // that refer to that meta-data file (recall that the counter data
    92  // file name incorporates the meta-data hash), and add the counter
    93  // data file to the appropriate pod.
    94  //
    95  // This process is complicated by the fact that we need to keep track
    96  // of directory indices for counter data files. Here is an example to
    97  // motivate:
    98  //
    99  //	directory 1:
   100  //
   101  // M1   covmeta.9bbf1777f47b3fcacb05c38b035512d6
   102  // C1   covcounters.9bbf1777f47b3fcacb05c38b035512d6.1677673.1662138360208416486
   103  // C2   covcounters.9bbf1777f47b3fcacb05c38b035512d6.1677637.1662138359974441782
   104  //
   105  //	directory 2:
   106  //
   107  // M2   covmeta.9bbf1777f47b3fcacb05c38b035512d6
   108  // C3   covcounters.9bbf1777f47b3fcacb05c38b035512d6.1677445.1662138360208416480
   109  // C4   covcounters.9bbf1777f47b3fcacb05c38b035512d6.1677677.1662138359974441781
   110  // M3   covmeta.a723844208cea2ae80c63482c78b2245
   111  // C5   covcounters.a723844208cea2ae80c63482c78b2245.3677445.1662138360208416480
   112  // C6   covcounters.a723844208cea2ae80c63482c78b2245.1877677.1662138359974441781
   113  //
   114  // In these two directories we have three meta-data files, but only
   115  // two are distinct, meaning that we'll wind up with two pods. The
   116  // first pod (with meta-file M1) will have four counter data files
   117  // (C1, C2, C3, C4) and the second pod will have two counter data files
   118  // (C5, C6).
   119  func collectPodsImpl(files []string, dirIndices []int, warn bool) []Pod {
   120  	metaRE := regexp.MustCompile(fmt.Sprintf(`^%s\.(\S+)$`, coverage.MetaFilePref))
   121  	mm := make(map[string]protoPod)
   122  	for _, f := range files {
   123  		base := filepath.Base(f)
   124  		if m := metaRE.FindStringSubmatch(base); m != nil {
   125  			tag := m[1]
   126  			// We need to allow for the possibility of duplicate
   127  			// meta-data files. If we hit this case, use the
   128  			// first encountered as the canonical version.
   129  			if _, ok := mm[tag]; !ok {
   130  				mm[tag] = protoPod{mf: f}
   131  			}
   132  			// FIXME: should probably check file length and hash here for
   133  			// the duplicate.
   134  		}
   135  	}
   136  	counterRE := regexp.MustCompile(fmt.Sprintf(coverage.CounterFileRegexp, coverage.CounterFilePref))
   137  	for k, f := range files {
   138  		base := filepath.Base(f)
   139  		if m := counterRE.FindStringSubmatch(base); m != nil {
   140  			tag := m[1] // meta hash
   141  			pid, err := strconv.Atoi(m[2])
   142  			if err != nil {
   143  				continue
   144  			}
   145  			if v, ok := mm[tag]; ok {
   146  				idx := -1
   147  				if dirIndices != nil {
   148  					idx = dirIndices[k]
   149  				}
   150  				fo := fileWithAnnotations{file: f, origin: idx, pid: pid}
   151  				v.elements = append(v.elements, fo)
   152  				mm[tag] = v
   153  			} else {
   154  				if warn {
   155  					warning("skipping orphaned counter file: %s", f)
   156  				}
   157  			}
   158  		}
   159  	}
   160  	if len(mm) == 0 {
   161  		if warn {
   162  			warning("no coverage data files found")
   163  		}
   164  		return nil
   165  	}
   166  	pods := make([]Pod, 0, len(mm))
   167  	for _, p := range mm {
   168  		sort.Slice(p.elements, func(i, j int) bool {
   169  			if p.elements[i].origin != p.elements[j].origin {
   170  				return p.elements[i].origin < p.elements[j].origin
   171  			}
   172  			return p.elements[i].file < p.elements[j].file
   173  		})
   174  		pod := Pod{
   175  			MetaFile:         p.mf,
   176  			CounterDataFiles: make([]string, 0, len(p.elements)),
   177  			Origins:          make([]int, 0, len(p.elements)),
   178  			ProcessIDs:       make([]int, 0, len(p.elements)),
   179  		}
   180  		for _, e := range p.elements {
   181  			pod.CounterDataFiles = append(pod.CounterDataFiles, e.file)
   182  			pod.Origins = append(pod.Origins, e.origin)
   183  			pod.ProcessIDs = append(pod.ProcessIDs, e.pid)
   184  		}
   185  		pods = append(pods, pod)
   186  	}
   187  	sort.Slice(pods, func(i, j int) bool {
   188  		return pods[i].MetaFile < pods[j].MetaFile
   189  	})
   190  	return pods
   191  }
   192  
   193  func warning(s string, a ...interface{}) {
   194  	fmt.Fprintf(os.Stderr, "warning: ")
   195  	fmt.Fprintf(os.Stderr, s, a...)
   196  	fmt.Fprintf(os.Stderr, "\n")
   197  }
   198
View as plain text