From 7ab37e5e10ef945f7f8603ff5ffe163b3deca9e9 Mon Sep 17 00:00:00 2001
From: Mirror <voice@magicalmirro.red>
Date: Fri, 26 Feb 2021 01:05:07 +0300
Subject: [PATCH] [WIP] typical filenames matcher

---
 main.go            |  3 ++-
 pixiv/filenames.go |  6 ++++++
 pixiv/new.go       | 23 +++++++++++++++++++++++
 3 files changed, 31 insertions(+), 1 deletion(-)
 create mode 100644 pixiv/filenames.go

diff --git a/main.go b/main.go
index 5e6879f..d13a263 100644
--- a/main.go
+++ b/main.go
@@ -24,6 +24,7 @@ type cmdlineT struct {
 	threads       int
 	setXAttrTags  bool
 	setTagsFile   string
+	testString    string
 }
 
 var (
@@ -63,7 +64,6 @@ func main() {
 			log.Fatal(err.Error())
 		}
 	}
-
 }
 
 func fetchFollows(userID string) (err error) {
@@ -159,5 +159,6 @@ func flags() {
 	flag.IntVar(&cmdline.threads, "threads", 1, "threads number")
 	flag.StringVar(&cmdline.setTagsFile, "set-xattr-file", "", "[WIP] fetch illustration's tags and write them to xattrs of the specified file. Filename SHOULD be in following formats: <illust_id>_p<page_number>.<file_ext>, illust_<illust_id>_*")
 	flag.BoolVar(&cmdline.setXAttrTags, "set-xattr", false, "also write tags to xattrs")
+	flag.StringVar(&cmdline.testString, "test-string", "", "used for testing purposes")
 	flag.Parse()
 }
diff --git a/pixiv/filenames.go b/pixiv/filenames.go
new file mode 100644
index 0000000..8177ddc
--- /dev/null
+++ b/pixiv/filenames.go
@@ -0,0 +1,6 @@
+package pixiv
+
+const (
+	// should have subexpressions matching only for IDs
+	typicalFilenamesRegex = "(?:(?:illust_)(?P<illust_id_a>[0-9]*)(?:_[0-9]{8}_[0-9]{6}\\.(?:jpg|png|ugoira)))|(?:(?P<illust_id_b>[0-9]*)(?:_p[0-9]*\\.(?:jpg|png|ugoira)))"
+)
diff --git a/pixiv/new.go b/pixiv/new.go
index 30dcd5b..a213d8d 100644
--- a/pixiv/new.go
+++ b/pixiv/new.go
@@ -5,6 +5,7 @@ import (
 	"log"
 	"net/http"
 	"os"
+	"regexp"
 
 	"h12.io/socks"
 )
@@ -20,6 +21,8 @@ type Pixiv struct {
 	logChannel      chan string
 	DownloadChannel chan Illust
 	setxattr        bool
+	nameMatcher     *regexp.Regexp
+	idSubexpNumbers []int
 }
 
 //New returns object with methods to access API functions
@@ -31,6 +34,7 @@ func New(cookies string, logFilePath string, threads int, xattrs bool) (p Pixiv)
 	p.ItemsPerRequest = 100
 	p.WorkDirectory = fmt.Sprintf("%s/Pictures/pixiv", os.Getenv("HOME"))
 	p.setxattr = xattrs
+	p.nameMatcher = regexp.MustCompile(typicalFilenamesRegex)
 	if len(logFilePath) > 0 {
 		logfile, err := os.OpenFile(logFilePath, os.O_APPEND, 664)
 		if err != nil {
@@ -71,3 +75,22 @@ func (p *Pixiv) SetProxy(proxy string) (err error) {
 	p.client.Transport = tr
 	return
 }
+
+// TypicalFilenamesMatcher matches typical pixiv's filenames. Returns ok if strings match and id extracted from string.
+func (p *Pixiv) TypicalFilenamesMatcher(name string) (ok bool, id string) {
+	ok = p.nameMatcher.MatchString(name)
+	if ok {
+		for subexpIndex, subexpName := range p.nameMatcher.SubexpNames() {
+			if subexpIndex == 0 {
+				// we don't need main expression here
+				continue
+			}
+			// try to read ID with each subexp and return on first match
+			id = p.nameMatcher.ReplaceAllString(name, fmt.Sprintf("${%s}", subexpName))
+			if len(id) > 0 {
+				break
+			}
+		}
+	}
+	return
+}