From 74795db477c7d5d03faeda159cdc58356f8f53fa Mon Sep 17 00:00:00 2001
From: Joseph Montanaro <jfmonty2@gmail.com>
Date: Wed, 28 Jul 2021 17:01:49 -0700
Subject: [PATCH] more refactoring

---
 .gitignore                   |  2 +-
 {src => data}/counts.txt     |  0
 {src => data}/dictionary.txt |  0
 passphrase.nimble            |  5 ++---
 src/passphrase.nim           |  2 +-
 src/passphrase.nims          |  1 +
 src/process.nim              | 38 +++++++++++++++++++++---------------
 7 files changed, 27 insertions(+), 21 deletions(-)
 rename {src => data}/counts.txt (100%)
 rename {src => data}/dictionary.txt (100%)
 create mode 100644 src/passphrase.nims

diff --git a/.gitignore b/.gitignore
index b0b6a2a..e222c89 100644
--- a/.gitignore
+++ b/.gitignore
@@ -1,2 +1,2 @@
 *.exe
-BNC/*
\ No newline at end of file
+data/BNC/*
\ No newline at end of file
diff --git a/src/counts.txt b/data/counts.txt
similarity index 100%
rename from src/counts.txt
rename to data/counts.txt
diff --git a/src/dictionary.txt b/data/dictionary.txt
similarity index 100%
rename from src/dictionary.txt
rename to data/dictionary.txt
diff --git a/passphrase.nimble b/passphrase.nimble
index 76fbfad..5df45eb 100644
--- a/passphrase.nimble
+++ b/passphrase.nimble
@@ -16,7 +16,6 @@ requires "nimcrypto >= 0.4.8"
 
 # Tasks n scripts
 
-import strutils
 
 proc runCmd(command: string, input = "", cache = ""): string =
   let (output, exitCode) = gorgeEx(command, input, cache)
@@ -29,10 +28,10 @@ proc runCmd(command: string, input = "", cache = ""): string =
 
 task(dictionary, "Generate dictionary from BNC XML files"):
   echo "Building dictionary"
-  let output = runCmd("nim c --run --threads:on -d:release -d:lto src/process.nim BNC/2554/download/Texts src/")
+  let output = runCmd("nim c --run --threads:on -d:release -d:lto src/process.nim data/BNC/2554/download/Texts src/")
   echo output.strip().splitlines()[^1]
 
 
 before(build):
-  if not fileExists("src/dictionary.txt"):
+  if not fileExists("data/dictionary.txt"):
     dictionaryTask()
diff --git a/src/passphrase.nim b/src/passphrase.nim
index d7d2de9..369f9a7 100644
--- a/src/passphrase.nim
+++ b/src/passphrase.nim
@@ -23,7 +23,7 @@ proc len(d: Dictionary): int =
 
 
 proc loadWords(): Dictionary =
-  for word in staticRead("dictionary.txt").strip().splitLines():
+  for word in staticRead("../data/dictionary.txt").strip().splitLines():
     let startIdx = result.words.len.uint32
     result.offsets.add(startIdx)
     result.words.add(word)
diff --git a/src/passphrase.nims b/src/passphrase.nims
new file mode 100644
index 0000000..99f2a9a
--- /dev/null
+++ b/src/passphrase.nims
@@ -0,0 +1 @@
+--d: release
\ No newline at end of file
diff --git a/src/process.nim b/src/process.nim
index c4badb7..a638cfd 100644
--- a/src/process.nim
+++ b/src/process.nim
@@ -27,6 +27,24 @@ proc save(wordCounts: CountTable; dictName, countName: string) =
       break
 
 
+type Config = object
+  srcPath: string
+  dstPath: string
+
+
+proc parseInput(): Config =
+  result.srcPath = r"../data/BNC/2554/download/Texts/"
+  result.dstPath = "."
+
+  if paramCount() > 0:
+    result.srcPath = paramStr(1)
+  if paramCount() > 1:
+    result.dstPath = paramStr(2)
+
+  if not dirExists(result.srcPath):
+    quit("Could not locate datafiles: directory " & result.srcPath & " does not exist.")
+
+
 var 
   threadResults: Channel[CountTable[string]]
   progress: Channel[int]
@@ -46,17 +64,10 @@ proc processFiles(filenames: seq[string]) =
 when isMainModule:
   let start = getMonoTime()
 
-  let basePath =
-    if paramCount() > 0:
-      paramStr(1)
-    else:
-      r"../BNC/2554/download/Texts/"
-
-  if not dirExists(basePath):
-    quit("Could not locate datafiles: directory " & basePath & " does not exist.")
+  let config = parseInput()
   
   var paths: seq[string]
-  for path in walkDirRec(basePath):
+  for path in walkDirRec(config.srcPath):
     if path.endsWith(".xml"):
       paths.add(path)
 
@@ -79,15 +90,10 @@ when isMainModule:
       counts.inc(word, count)
 
   counts.sort()
-  let outPath = 
-    if paramCount() > 1:
-      paramStr(2)
-    else:
-      "."
   
   let
-    dPath = joinPath(outPath, "dictionary.txt")
-    cPath = joinPath(outPath, "counts.txt")
+    dPath = joinPath(config.dstPath, "dictionary.txt")
+    cPath = joinPath(config.dstPath, "counts.txt")
   save(counts, dPath, cPath)
 
   echo "Done. Finished in ", (getMonoTime() - start).inMilliseconds.float / 1000, " seconds."