Merge pull request #56 from jiro4989/feature/#15-file

add file provider #15

README.rst

@@ -79,15 +79,15 @@ TODO list.
 
 - [ ] base
 - [o] faker.provider.address
-- [ ] faker.provider.automotive
+- [o] faker.provider.automotive
 - [o] faker.provider.bank
 - [ ] faker.provider.barcode
 - [ ] faker.provider.color
 - [o] faker.provider.company
 - [ ] faker.provider.credit_card
 - [o] faker.provider.currency
 - [ ] faker.provider.date_time
-- [ ] faker.provider.file
+- [o] faker.provider.file
 - [ ] faker.provider.geo
 - [ ] faker.provider.internet
 - [ ] faker.provider.isbn

faker.nimble

@@ -1,6 +1,6 @@
 # Package
 
-version       = "0.10.0"
+version       = "0.11.0"
 author        = "jiro4989"
 description   = "faker is a Nim package that generates fake data for you."
 license       = "MIT"
@@ -30,7 +30,7 @@ proc addGeneratedText(lines: var seq[string]) =
 proc readPublicProcs(file: string): seq[string] =
   readFile(file)
     .split("\n")
-    .filterIt(it.startsWith("proc") and "*(f: Faker)" in it)
+    .filterIt(it.startsWith("proc") and "*(f: Faker" in it)
 
 proc readImplementedLocales(dir, provider: string): seq[string] =
   for path in listFiles(dir):
@@ -61,7 +61,7 @@ proc genProviderIndexFile(provider: string) =
     lines.add(&"    let f = newFaker()")
     let arg2 =
       if 1 < args.split(",").len:
-        args.replace("f: Faker, ", "")
+        args.replace("f: Faker, ", "").replace("(", "").replace(")", "")
       else:
         ""
     lines.add(&"    echo f.{procName}({arg2})")

src/faker/provider.nim

@@ -3,5 +3,5 @@
 # See 'faker.nimble'.                             #
 # ----------------------------------------------- #
 
-import provider/[address, automotive, bank, company, currency, job, person, phone_number, user_agent]
-export address, automotive, bank, company, currency, job, person, phone_number, user_agent
+import provider/[address, automotive, bank, company, currency, file, job, person, phone_number, user_agent, word]
+export address, automotive, bank, company, currency, file, job, person, phone_number, user_agent, word

src/faker/provider/file.nim

@@ -0,0 +1,68 @@
+# ----------------------------------------------- #
+# This module was generated by 'nimble genProvs'. #
+# See 'faker.nimble'.                             #
+# ----------------------------------------------- #
+
+import ../base
+import file/[file_en_US]
+export base
+
+proc fileExtension*(f: Faker, category=""): string =
+  ## Generates random fileExtension.
+  runnableExamples:
+    let f = newFaker()
+    echo f.fileExtension(category="")
+
+  case f.locale
+  of "en_US": file_en_US.fileExtension(f, category="")
+  else: file_en_US.fileExtension(f, category="")
+
+proc fileName*(f: Faker, category="", extension=""): string =
+  ## Generates random fileName.
+  runnableExamples:
+    let f = newFaker()
+    echo f.fileName(category="", extension="")
+
+  case f.locale
+  of "en_US": file_en_US.fileName(f, category="", extension="")
+  else: file_en_US.fileName(f, category="", extension="")
+
+proc filePath*(f: Faker, depth=1, category="", extension=""): string =
+  ## Generates random filePath.
+  runnableExamples:
+    let f = newFaker()
+    echo f.filePath(depth=1, category="", extension="")
+
+  case f.locale
+  of "en_US": file_en_US.filePath(f, depth=1, category="", extension="")
+  else: file_en_US.filePath(f, depth=1, category="", extension="")
+
+proc mimeType*(f: Faker, category=""): string =
+  ## Generates random mimeType.
+  runnableExamples:
+    let f = newFaker()
+    echo f.mimeType(category="")
+
+  case f.locale
+  of "en_US": file_en_US.mimeType(f, category="")
+  else: file_en_US.mimeType(f, category="")
+
+proc unixDevice*(f: Faker, prefix=""): string =
+  ## Generates random unixDevice.
+  runnableExamples:
+    let f = newFaker()
+    echo f.unixDevice(prefix="")
+
+  case f.locale
+  of "en_US": file_en_US.unixDevice(f, prefix="")
+  else: file_en_US.unixDevice(f, prefix="")
+
+proc unixPartition*(f: Faker, prefix=""): string =
+  ## Generates random unixPartition.
+  runnableExamples:
+    let f = newFaker()
+    echo f.unixPartition(prefix="")
+
+  case f.locale
+  of "en_US": file_en_US.unixPartition(f, prefix="")
+  else: file_en_US.unixPartition(f, prefix="")

src/faker/provider/file/file_en_US.nim

@@ -0,0 +1,239 @@
+from strformat import `&`
+from sequtils import toSeq
+import tables
+
+include interfaces
+import ../util
+import ../word
+
+const
+  applicationMimeTypes = @[
+    "application/atom+xml",  # Atom feeds
+    "application/ecmascript",
+    # ECMAScript/JavaScript; Defined in RFC 4329 (equivalent to
+    # application/javascript but with stricter processing rules)
+    "application/EDI-X12",  # EDI X12 data; Defined in RFC 1767
+    "application/EDIFACT",  # EDI EDIFACT data; Defined in RFC 1767
+    "application/json",  # JavaScript Object Notation JSON; Defined in RFC 4627
+    # ECMAScript/JavaScript; Defined in RFC 4329 (equivalent to
+    # application/ecmascript
+    "application/javascript",
+    #   but with looser processing rules) It is not accepted in IE 8
+    #   or earlier - text/javascript is accepted but it is defined as obsolete in RFC 4329.
+    #   The "type" attribute of the <script> tag in HTML5 is optional and in practice
+    #   omitting the media type of JavaScript programs is the most interoperable
+    #   solution since all browsers have always assumed the correct
+    #   default even before HTML5.
+    "application/octet-stream",
+    # Arbitrary binary data.[6] Generally speaking this type identifies files that are not associated with
+    # a specific application. Contrary to past assumptions by software packages such as Apache this is not
+    # a type that should be applied to unknown files. In such a case, a server or application should not indicate
+    # a content type, as it may be incorrect, but rather, should omit the type in order to allow the recipient
+    # to guess the type.[7]
+    "application/ogg",  # Ogg, a multimedia bitstream container format; Defined in RFC 5334
+    "application/pdf",  # Portable Document Format, PDF has been in use for document exchange
+    #   on the Internet since 1993; Defined in RFC 3778
+    "application/postscript",  # PostScript; Defined in RFC 2046
+    "application/rdf+xml",  # Resource Description Framework; Defined by RFC 3870
+    "application/rss+xml",  # RSS feeds
+    "application/soap+xml",  # SOAP; Defined by RFC 3902
+    # Web Open Font Format; (candidate recommendation; use application/x-font-woff
+    "application/font-woff",
+    #   until standard is official)
+    "application/xhtml+xml",  # XHTML; Defined by RFC 3236
+    "application/xml-dtd",  # DTD files; Defined by RFC 3023
+    "application/xop+xml",  # XOP
+    "application/zip",  # ZIP archive files; Registered[8]
+    "application/gzip",         # Gzip, Defined in RFC 6713
+  ]
+
+  audioMimeTypes = @[
+    "audio/basic",  # mulaw audio at 8 kHz, 1 channel; Defined in RFC 2046
+    "audio/L24",  # 24bit Linear PCM audio at 8-48 kHz, 1-N channels; Defined in RFC 3190
+    "audio/mp4",  # MP4 audio
+    "audio/mpeg",  # MP3 or other MPEG audio; Defined in RFC 3003
+    "audio/ogg",  # Ogg Vorbis, Speex, Flac and other audio; Defined in RFC 5334
+    "audio/vorbis",  # Vorbis encoded audio; Defined in RFC 5215
+    # RealAudio; Documented in RealPlayer Help[9]
+    "audio/vnd.rn-realaudio",
+    "audio/vnd.wave",  # WAV audio; Defined in RFC 2361
+    "audio/webm",               # WebM open media format
+  ]
+
+  imageMimeTypes = @[
+    "image/gif",  # GIF image; Defined in RFC 2045 and RFC 2046
+    "image/jpeg",  # JPEG JFIF image; Defined in RFC 2045 and RFC 2046
+    "image/pjpeg",
+    # JPEG JFIF image; Associated with Internet Explorer; Listed in ms775147(v=vs.85) - Progressive JPEG,
+    # initiated before global browser support for progressive JPEGs (Microsoft and Firefox).
+    # Portable Network Graphics; Registered,[10] Defined in RFC 2083
+    "image/png",
+    "image/svg+xml",  # SVG vector image; Defined in SVG Tiny 1.2 Specification Appendix M
+    # Tag Image File Format (only for Baseline TIFF); Defined in RFC 3302
+    "image/tiff",
+    "image/vnd.microsoft.icon",  # ICO image; Registered[11]
+  ]
+
+  messageMimeTypes = @[
+    "message/http",  # Defined in RFC 2616
+    "message/imdn+xml",  # IMDN Instant Message Disposition Notification; Defined in RFC 5438
+    "message/partial",  # Email; Defined in RFC 2045 and RFC 2046
+    # Email; EML files, MIME files, MHT files, MHTML files; Defined in RFC
+    # 2045 and RFC 2046
+    "message/rfc822",
+  ]
+
+  modelMimeTypes = @[
+    "model/example",  # Defined in RFC 4735
+    "model/iges",  # IGS files, IGES files; Defined in RFC 2077
+    "model/mesh",  # MSH files, MESH files; Defined in RFC 2077, SILO files
+    "model/vrml",  # WRL files, VRML files; Defined in RFC 2077
+    # X3D ISO standard for representing 3D computer graphics, X3DB binary
+    # files
+    "model/x3d+binary",
+    "model/x3d+vrml",  # X3D ISO standard for representing 3D computer graphics, X3DV VRML files
+    "model/x3d+xml",  # X3D ISO standard for representing 3D computer graphics, X3D XML files
+  ]
+
+  multipartMimeTypes = @[
+    "multipart/mixed",  # MIME Email; Defined in RFC 2045 and RFC 2046
+    "multipart/alternative",  # MIME Email; Defined in RFC 2045 and RFC 2046
+    # MIME Email; Defined in RFC 2387 and used by MHTML (HTML mail)
+    "multipart/related",
+    "multipart/form-data",  # MIME Webform; Defined in RFC 2388
+    "multipart/signed",  # Defined in RFC 1847
+    "multipart/encrypted",  # Defined in RFC 1847
+  ]
+
+  textMimeTypes = @[
+    "text/cmd",  # commands; subtype resident in Gecko browsers like Firefox 3.5
+    "text/css",  # Cascading Style Sheets; Defined in RFC 2318
+    "text/csv",  # Comma-separated values; Defined in RFC 4180
+    "text/html",  # HTML; Defined in RFC 2854
+    "text/javascript",
+    # (Obsolete): JavaScript; Defined in and obsoleted by RFC 4329 in order to discourage its usage in favor of
+    # application/javascript. However, text/javascript is allowed in HTML 4 and 5 and, unlike
+    # application/javascript, has cross-browser support. The "type" attribute of the <script> tag in HTML5 is
+    # optional and there is no need to use it at all since all browsers have always assumed the correct default
+    # (even in HTML 4 where it was required by the specification).
+    "text/plain",  # Textual data; Defined in RFC 2046 and RFC 3676
+    "text/vcard",  # vCard (contact information); Defined in RFC 6350
+    "text/xml",  # Extensible Markup Language; Defined in RFC 3023
+  ]
+
+  videoMimeTypes = @[
+    "video/mpeg",  # MPEG-1 video with multiplexed audio; Defined in RFC 2045 and RFC 2046
+    "video/mp4",  # MP4 video; Defined in RFC 4337
+    # Ogg Theora or other video (with audio); Defined in RFC 5334
+    "video/ogg",
+    "video/quicktime",  # QuickTime video; Registered[12]
+    "video/webm",  # WebM Matroska-based open media format
+    "video/x-matroska",  # Matroska open media format
+    "video/x-ms-wmv",  # Windows Media Video; Documented in Microsoft KB 288102
+    "video/x-flv",  # Flash video (FLV files)
+  ]
+
+  mimeTypes = {
+    "application": applicationMimeTypes,
+    "audio": audioMimeTypes,
+    "image": imageMimeTypes,
+    "message": messageMimeTypes,
+    "model": modelMimeTypes,
+    "multipart": multipartMimeTypes,
+    "text": textMimeTypes,
+    "video": videoMimeTypes,
+  }.toTable
+
+  audioFileExtensions = @[
+    "flac",
+    "mp3",
+    "wav",
+  ]
+
+  imageFileExtensions = @[
+    "bmp",
+    "gif",
+    "jpeg",
+    "jpg",
+    "png",
+    "tiff",
+  ]
+
+  textFileExtensions = @[
+    "css",
+    "csv",
+    "html",
+    "js",
+    "json",
+    "txt",
+  ]
+
+  videoFileExtensions = @[
+    "mp4",
+    "avi",
+    "mov",
+    "webm",
+  ]
+
+  officeFileExtensions = @[
+    "doc",  # legacy MS Word
+    "docx",  # MS Word
+    "xls",  # legacy MS Excel
+    "xlsx",  # MS Excel
+    "ppt",  # legacy MS PowerPoint
+    "pptx",  # MS PowerPoint
+    "odt",  # LibreOffice document
+    "ods",  # LibreOffice spreadsheet
+    "odp",  # LibreOffice presentation
+    "pages",  # Apple Pages
+    "numbers",  # Apple Numbers
+    "key",  # Apple Keynote
+    "pdf",  # Portable Document Format
+  ]
+
+  fileExtensions = {
+    "audio": audioFileExtensions,
+    "image": imageFileExtensions,
+    "office": officeFileExtensions,
+    "text": textFileExtensions,
+    "video": videoFileExtensions,
+  }.toTable
+
+  unixDevicePrefixes = @["sd", "vd", "xvd"]
+
+proc filePath*(f: Faker, depth=1, category="", extension=""): string =
+  let filename = f.fileName(category=category, extension=extension)
+  result = &"/{filename}"
+  for _ in 0..<depth:
+    let word = f.word()
+    result = &"/{word}{result}"
+
+proc fileName*(f: Faker, category="", extension=""): string =
+  let filename = f.word()
+  let extension =
+    if extension == "": f.fileExtension(category=category)
+    else: extension
+  result = &"{filename}.{extension}"
+
+proc fileExtension*(f: Faker, category=""): string =
+  let category =
+    if category == "": f.rand.sample(toSeq(fileExtensions.keys))
+    else: category
+  result = f.rand.sample(fileExtensions[category])
+
+proc mimeType*(f: Faker, category=""): string =
+  let category =
+    if category == "": f.rand.sample(toSeq(mimeTypes.keys))
+    else: category
+  result = f.rand.sample(mimeTypes[category])
+
+proc unixDevice*(f: Faker, prefix=""): string =
+  let prefix =
+    if prefix == "": f.rand.sample(unixDevicePrefixes)
+    else: prefix
+  let suffix = f.rand.sample(asciiLowerCase)
+  result = &"/dev/{prefix}{suffix}"
+
+proc unixPartition*(f: Faker, prefix=""): string =
+  let path = f.unixDevice(prefix=prefix)
+  result = path & $f.rand.rand(0..9)

src/faker/provider/file/interfaces.nim

@@ -0,0 +1,16 @@
+# interfaces はプロシージャの定義のみをまとめたモジュール。
+# C言語でいうヘッダファイルのように使用する。
+#
+# このモジュールをincludeしたモジュールは、このモジュールに定義しているプロシー
+# ジャを全て実装しないと、コンパイルが通らない。
+
+import random
+import ../../base
+export base
+
+proc fileExtension*(f: Faker, category=""): string
+proc fileName*(f: Faker, category="", extension=""): string
+proc filePath*(f: Faker, depth=1, category="", extension=""): string
+proc mimeType*(f: Faker, category=""): string
+proc unixDevice*(f: Faker, prefix=""): string
+proc unixPartition*(f: Faker, prefix=""): string

src/faker/provider/util.nim

@@ -3,7 +3,8 @@ from strutils import join, align
 from math import `^`
 
 const
-  asciiUpperCase = {'A'..'Z'}
+  asciiUpperCase* = {'A'..'Z'}
+  asciiLowerCase* = {'a'..'z'}
 
 proc formatNumbers*(r: var Rand, format: string): string =
   ## Replace '#' to random number.