Updates for Nim 2.0

build.gradle

@@ -3,17 +3,18 @@ apply plugin: "application"
 apply plugin: "maven"
 
 group = "com.jdblabs"
-version = "1.4.2"
+version = "1.4.4"
 mainClassName = "com.jdblabs.file.treediff.TreeDiff"
 
 repositories {
     mavenLocal()
-    mavenCentral() }
+    mavenCentral()
+    maven { url "http://mvn.jdb-labs.com/repo" } }
 
 dependencies {
-    compile 'org.codehaus.groovy:groovy-all:2.4.3'
-	compile 'com.jdbernard:jdb-util:3.8'
-	compile 'commons-codec:commons-codec:1.10'
+    compile localGroovy()
+    compile 'com.jdbernard:jdb-util:4.+'
+    compile 'commons-codec:commons-codec:1.10'
     compile 'com.fasterxml.jackson.core:jackson-databind:2.4.4'
 
     testCompile 'junit:junit:4.12'

src/main/groovy/com/jdblabs/file/treediff/TreeDiff.groovy

@@ -8,7 +8,7 @@ import org.apache.commons.codec.digest.DigestUtils
 
 public class TreeDiff {
 
-    public static final String VERSION = "1.4.2"
+    public static final String VERSION = "1.4.3"
 
     private ObjectMapper objectMapper = new ObjectMapper()
     private PrintStream stdout

src/main/nim/cliconstants.nim

@@ -0,0 +1,63 @@
+const VERSION* = "2.0.1"
+
+const USAGE* = """
+Usage:
+  treediff <left> [<right>] [options]
+  treediff (-h | --help)
+  treediff (-V | --version)
+
+  <left> and <right> represent paths to directory roots to be compared. If one
+  of these paths points to a file instead of a directory, treediff assumes that
+  the file represents a saved directory analysis to be loaded in place of a
+  directory to compare. For example:
+
+      treediff /path/to/dir /path/to/output.json
+
+  will analyze the directory tree at '/path/to/dir' to create the left-side
+  analysis and load a pre-existing analysis from '/path/to/output.json' as the
+  right-side analysis.
+
+Options:
+  -h --help       Show this usage information.
+  -V --version    Show the program version.
+  -v --verbose    Enable verbose output.
+  -q --quiet      Suppress all output and error messages except for the
+                  progress indicator.
+  -Q --very-quiet Suppress all output and error messages includeing the
+                  progress indicator.
+
+  -1 --save-left <left_out>   Save the left analysis to <left_out> (will be
+                              formatted as JSON)
+  -2 --save-right <right_out> Save the right analysis to <right_out> (will be
+                              formatted as JSON)
+
+  -s --same
+  -S --exclude-same
+
+      Show or hide information about files which are the same in both trees.
+
+  -c --content-mismatch
+  -C --exclude-content-mismatch
+
+      Show or hide information about files whose relative paths are the same
+      in both trees but whose contents differ.
+
+  -p --path-mismatch
+  -P --exclude-path-mismatch
+
+      Show or hide information about files whose contents are the same in both
+      trees but whose relative paths differ.
+
+  -l --left-only
+  -L --exclude-left-only
+
+      Show or hide information about files which are found only in the left
+      tree.
+
+  -r --right-only
+  -R --exclude-right-only
+
+      Show or hide information about files which are found only in the right
+      tree.
+
+"""

src/main/nim/incremental_md5.nim

@@ -1,37 +1,33 @@
-import md5
-import os
+import std/streams
+import checksums/md5
 
 proc fileToMD5*(filename: string) : string =
-  
-  const blockSize: int = 8192
+
+  const blockSize: int = 8192 # read files in 8KB chunnks
   var
     c: MD5Context
     d: MD5Digest
-    f: File
-    bytesRead: int = 0
-    buffer: array[blockSize, char]
-    byteTotal: int = 0
-  
+    fs: FileStream
+    buffer: string
+
   #read chunk of file, calling update until all bytes have been read
   try:
-    f = open(filename)
-    
+    fs = filename.open.newFileStream
+
     md5Init(c)
-    bytesRead = f.readBuffer(buffer.addr, blockSize)
-    
-    while bytesRead > 0:
-      byteTotal += bytesRead
-      md5Update(c, buffer, bytesRead)
-      bytesRead = f.readBuffer(buffer.addr, blockSize)
-    
+    buffer = fs.readStr(blockSize)
+
+    while buffer.len > 0:
+      md5Update(c, buffer.cstring, buffer.len)
+      buffer = fs.readStr(blockSize)
+
     md5Final(c, d)
-  
-  except IOError:
-    echo("File not found.")
+
+  except IOError: echo("File not found.")
   finally:
-    if f != nil:
-      close(f)
-  
+    if fs != nil:
+      close(fs)
+
   result = $d
 
 when isMainModule:

src/main/nim/treediff.nim

@@ -3,9 +3,11 @@
 ##
 ## Utility to compare the file contents of two directory trees.
 
-import os, tables, streams, sequtils, strutils, docopt, marshal
+import std/[json, jsonutils, os, tables, sequtils, strutils]
+import docopt
 import incremental_md5, console_progress
 
+import ./cliconstants
 
 
 type
@@ -20,16 +22,17 @@ type
 proc newProgressWrapper*(outFile = stdout, verbosity = normal): ProgressWrapper =
   ## Create a new ProgressWrapper for the given verbosity.
   if verbosity > very_quiet:
-    result = (impl: newProgress(outFile, 0), verbosity: verbosity)
+    result = (impl: newProgress(0, outFile), verbosity: verbosity)
   else: result = (impl: nil, verbosity: verbosity)
 
 proc init(p: ProgressWrapper, root: string, fileCount: int): void =
-  if p.verbosity == normal: echo "-- ", root.expandFilename
+  if p.verbosity == normal:
+      echo "-- ", root.expandFilename, "\L   ", fileCount, " files"
   if p.verbosity > very_quiet: p.impl.setMax(fileCount)
 
 proc update(p: ProgressWrapper, count: int, file: string): void =
   if p.verbosity > very_quiet:
-    p.impl.updateProgress(count, file[(file.high - 15)..file.high])
+    p.impl.updateProgress(count, file[max(file.high - 15, 0)..file.high])
 
 proc finish(p: ProgressWrapper): void =
   if p.verbosity > very_quiet:
@@ -62,99 +65,121 @@ proc getRelPath(ancestor, child: string): string =
   # build a relative path without backtracking.
   if idx != ancestorPath.len: return ""
   return foldl(@["."] & childPath[idx..childPath.high], joinPath(a, b))
-  
+
 
 
 
 type
-  FileEntry* = tuple[relPath: string, checksum: string]
+  FileEntry* = ref tuple[relPath: string, checksum: string]
     ## Data about one file that has been analyzed
 
   DirAnalysis* = ## Analysis data about one directory tree.
-    tuple[allEntries: seq[ref FileEntry],
-          byRelPath: ref Table[string, ref FileEntry],
-          byChecksum: ref Table[string, seq[ref FileEntry]]]
+    tuple[allEntries: seq[FileEntry],
+          byRelPath: TableRef[string, FileEntry],
+          byChecksum: TableRef[string, seq[FileEntry]]]
 
 
   DisplayOptions = tuple[left, right, same, content, path: bool]
     ## Consolidated description of which types of results to display.
 
+func `$`(f: FileEntry): string = f.checksum & ": " & f.relPath
+
+proc getOrFail(n: JsonNode, key: string, objName: string = ""): JsonNode =
+  ## convenience method to get a key from a JObject or raise an exception
+  if not n.hasKey(key): raise newException(Exception, objName & " missing key '" & key & "'")
+  return n[key]
+
+proc getIfExists(n: JsonNode, key: string): JsonNode =
+  ## convenience method to get a key from a JObject or return null
+  result = if n.hasKey(key): n[key]
+           else: newJNull()
+
+func parseFileEntry(n: JsonNode): FileEntry =
+  result = new(FileEntry)
+  result.relPath = n.getOrFail("relPath").getStr
+  result.checksum = n.getOrFail("checksum").getStr
+
+func initDirAnalysis(): DirAnalysis =
+  (allEntries: @[],
+   byRelPath: newTable[string, FileEntry](),
+   byChecksum: newTable[string, seq[FileEntry]]())
+
+func indexEntries(da: var DirAnalysis) =
+  for e in da.allEntries:
+    da.byRelPath[e.relPath] = e
+    if not da.byChecksum.hasKey(e.checksum):
+      da.byChecksum[e.checksum] = newSeq[FileEntry]()
+    da.byChecksum[e.checksum].add(e)
+
 proc analyzeDir*(root: string, progress: ProgressWrapper): DirAnalysis =
   ## Inspect a directory and analyze all files, noting their relative paths and
   ## checksum of their contents.
   let fileCount = countFiles(root)
 
-  progress.init(root, fileCount)
+  progress.init(root, fileCount + 10)
 
-  result = (allEntries: @[],
-            byRelPath: newTable[string, ref FileEntry](),
-            byChecksum: newTable[string, seq[ref FileEntry]]())
+  result = initDirAnalysis()
 
   var count = 0
   for file in walkDirRec(root):
-
-    # Compute checksum
     let md5sum = fileToMd5(file)
-    var fileEntry: ref FileEntry = new(ref FileEntry)
-    fileEntry[] = (relPath: getRelPath(root, file), checksum: md5sum )
 
-    # Add to allEntries list, byRelPath table, and byChecksum table
+    var fileEntry: FileEntry = new(FileEntry)
+    fileEntry[] = (relPath: getRelPath(root, file), checksum: md5sum)
     result.allEntries.add(fileEntry)
-    result.byRelPath[fileEntry.relPath] = fileEntry
-
-    if not result.byChecksum.hasKey(fileEntry.relPath):
-      result.byChecksum[fileEntry.checksum] = newSeq[ref FileEntry]()
-
-    result.byChecksum[fileEntry.checksum].add(fileEntry)
 
     progress.update(count, file)
     count += 1
 
+  result.indexEntries
+  count += 10
   progress.finish()
 
-proc loadAnalysis*(path: string, analysis: var DirAnalysis) =
+proc loadAnalysis*(path: string): DirAnalysis =
   ## Load a previously performed directory analysis.
-  let inStream: Stream = newFileStream(path, fmRead)
-  load(inStream, analysis)
+  let allEntriesJson = parseJson(readFile(path))
+  result = initDirAnalysis()
+  result.allEntries = toSeq(items(allEntriesJson)).map(parseFileEntry)
+  result.indexEntries
 
 proc saveAnalysis*(path: string, analysis: DirAnalysis): void =
   ## Save a completed analysis.
-  let outStream = newFileStream(path, fmWrite)
-  store(outStream, analysis)
+  writeFile(path, $(analysis.allEntries.toJson))
 
-proc intersection*(left, right: DirAnalysis): seq[ref FileEntry] =
+proc intersection*(left, right: DirAnalysis): seq[FileEntry] =
   ## Find all ``FileEntry`` that are the same on both sides: matching contents
   ## and paths.
-  return left.allEntries.filter do (item: ref FileEntry) -> bool:
+  return left.allEntries.filter do (item: FileEntry) -> bool:
     if not right.byRelPath.hasKey(item.relPath): return false
     let match = right.byRelPath[item.relPath]
+    if match == nil: return false
     return item.checksum == match.checksum
 
-proc difference*(left, right: DirAnalysis): seq[ref FileEntry] =
+proc difference*(left, right: DirAnalysis): seq[FileEntry] =
   ## Find all ``FileEntry`` that are present in the left but not present in
   ## the right.
-  return left.allEntries.filter do (item: ref FileEntry) -> bool:
+  return left.allEntries.filter do (item: FileEntry) -> bool:
     return not right.byRelPath.hasKey(item.relPath) and
            not right.byChecksum.hasKey(item.checksum)
 
-proc `*`*(left, right: DirAnalysis): seq[ref FileEntry] {.inline.} =
+proc `*`*(left, right: DirAnalysis): seq[FileEntry] {.inline.} =
   ## Alias for `intersection(left, right) <#intersection>`_
   return intersection(left, right)
 
-proc `-`*(left, right: DirAnalysis): seq[ref FileEntry] {.inline.} =
+proc `-`*(left, right: DirAnalysis): seq[FileEntry] {.inline.} =
   ## Alias for `difference(left, right) <#difference>`_
   return difference(left, right)
 
 proc samePathDifferentContents*(left, right: DirAnalysis): seq[string] =
   ## Find all ``FileEntry`` that have the same paths in both trees but whose
   ## contents differ.
-  let matchingEntries = left.allEntries.filter do (item: ref FileEntry) -> bool:
+  let matchingEntries = left.allEntries.filter do (item: FileEntry) -> bool:
     if not right.byRelPath.hasKey(item.relPath): return false
     let match = right.byRelPath[item.relPath]
     return item.checksum != match.checksum
-  return matchingEntries.map(proc(item: ref FileEntry): string = return item.relPath)
+  return matchingEntries.map(proc(item: FileEntry): string = return item.relPath)
 
-proc sameContentsDifferentPaths*(left, right: DirAnalysis): seq[tuple[left, right: ref FileEntry]] =
+proc sameContentsDifferentPaths*(left, right: DirAnalysis): seq[tuple[left, right: FileEntry]] =
   ## Find all ``FileEntry`` whose contents are the same in both trees but
   ## which are located at differenc paths.
   result = @[]
@@ -169,69 +194,7 @@ when isMainModule:
     stderr.writeLine("treediff: " & error)
     quit(QuitFailure)
 
-  let doc = """
-Usage:
-  treediff <left> [<right>] [options]
-  treediff (-h | --help)
-  treediff (-V | --version)
-
-  <left> and <right> represent paths to directory roots to be compared. If one
-  of these paths points to a file instead of a directory, treediff assumes that
-  the file represents a saved directory analysis to be loaded in place of a
-  directory to compare. For example:
-
-      treediff /path/to/dir /path/to/output.json
-
-  will analyze the directory tree at '/path/to/dir' to create the left-side
-  analysis and load a pre-existing analysis from '/path/to/output.json' as the
-  right-side analysis.
-  
-Options:
-  -h --help       Show this usage information.
-  -V --version    Show the program version.
-  -v --verbose    Enable verbose output.
-  -q --quiet      Suppress all output and error messages except for the
-                  progress indicator.
-  -Q --very-quiet Suppress all output and error messages includeing the
-                  progress indicator.
-
-  -1 --save-left <left_out>   Save the left analysis to <left_out> (will be
-                              formatted as JSON)
-  -2 --save-right <right_out> Save the right analysis to <right_out> (will be
-                              formatted as JSON)
-
-  -s --same
-  -S --exclude-same
-
-      Show or hide information about files which are the same in both trees.
-
-  -c --content-mismatch   
-  -C --exclude-content-mismatch
-
-      Show or hide information about files whose relative paths are the same
-      in both trees but whose contents differ.
-
-  -p --path-mismatch
-  -P --exclude-path-mismatch
-
-      Show or hide information about files whose contents are the same in both
-      trees but whose relative paths differ.
-
-  -l --left-only
-  -L --exclude-left-only
-
-      Show or hide information about files which are found only in the left
-      tree.
-
-  -r --right-only
-  -R --exclude-right-only
-
-      Show or hide information about files which are found only in the right
-      tree.
-
-"""
-
-  let args = docopt(doc, version = "treediff v1.4.2")
+  let args = docopt(USAGE, version = "treediff " & VERSION)
 
   var verbosity = normal
   if args["--quiet"]: verbosity = quiet
@@ -252,7 +215,7 @@ Options:
     if fileInfo.kind == pcDir:
       return analyzeDir(path, progressWrapper)
     elif fileInfo.kind == pcFile:
-      loadAnalysis(path, result)
+      result = loadAnalysis(path)
     else:
       quitWithError($path & ": is not a file or directory")
 
@@ -262,8 +225,8 @@ Options:
 
   if not args["<right>"]:
     rightAnalysis = (allEntries: @[],
-                 byRelPath: newTable[string, ref FileEntry](),
-                 byChecksum: newTable[string, seq[ref FileEntry]]())
+                 byRelPath: newTable[string, FileEntry](),
+                 byChecksum: newTable[string, seq[FileEntry]]())
   else:
     var rightPath: string = $args["<right>"]
     rightAnalysis = loadPath(rightPath)
@@ -274,7 +237,7 @@ Options:
 
   if args["--save-right"] and rightAnalysis.allEntries.len > 0:
     saveAnalysis($args["--save-right"], rightAnalysis)
-  
+
   # Parse filter options
   var displayOptions: DisplayOptions = (
     left: false, right: false, same: false, content: false, path: false)
@@ -319,4 +282,3 @@ Options:
     if displayOptions.right:
       let rightOnly = rightAnalysis - leftAnalysis
       for fe in rightOnly: echo "right only: ", fe.relPath
-

treediff.nimble

@@ -1,5 +1,5 @@
 # Package
-version       = "1.4.2"
+version       = "2.0.1"
 author        = "Jonathan Bernard (jdb@jdb-labs.com)"
 description   = "Utility to generate diffs of full directory trees."
 license       = "BSD"
@@ -7,4 +7,10 @@ bin           = @["treediff"]
 srcDir        = "src/main/nim"
 
 # Dependencies
-requires: @["nim >= 0.13.0", "docopt >= 0.1.0", "console_progress >= 1.2.1"]
+requires: @["nim >= 2.0.0", "docopt == 0.7.1", "checksums"]
+
+# Dependencies from git.jdb-software.com/jdb/nim-packages
+requires: @["console_progress >= 1.2.2", "update_nim_package_version"]
+
+task updateVersion, "Update the version of this package.":
+  exec "update_nim_package_version treediff 'src/main/nim/cliconstants.nim'"