xml_col_finder 0.1.1 → 0.2.3

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 45d703949948a3a22ed318a5297cf621eecd664e842ac1f654c497018e4b9ec1
-  data.tar.gz: f46900d22268f63f35e525bdc70156e0cf9bf68da7faf292facbe682c9094bc8
+  metadata.gz: 4229c7009179e903163ef10f64f0917a7f67e29d588172d184bed94f716123cf
+  data.tar.gz: f543fcbacf5ca517995d28fc6050671b48f18c8ae9df539aad6133486d7b74f5
 SHA512:
-  metadata.gz: c5da937ee341a1b26a587113633c51912c7f61f7409eebd0197c241860c799c266be394ed247009cbb6158cf4ca5f3de45ed4a46380b49736d25cecf86828f52
-  data.tar.gz: 2bfc5dee7af91f61327d71fafe91a73ea0e6060e8953dd193e6a9e1b6f839db9a0d33574cc3bb76f2c42824817e47d135edd866b1fb76108496c50d8ce4f7366
+  metadata.gz: 22cd4abb94f055f100d0f56993de883fa87d1bcb2e59ef9edaf0f3f46846f49aa2c73f7e0a38f4779d40314ad5a93bfe91b22738d347bf5866afc1b2ed58c105
+  data.tar.gz: 10c48be3a91d355aa826ebaf1e5b6329e2f3ec7acabc795e8800eb4758fe0abed45cfaee761a91190f72b4407e7b0a0a9accc19b563ca770057a98aa7dfb148b

data/lib/xml_col_finder.rb

@@ -12,10 +12,10 @@ class XMLColFinder
   def initialize(s, debug: false)
 
     @debug = debug
-    doc = Rexle.new(s)
+    @doc = Rexle.new(s)
 
     a = []
-    doc.root.each_recursive do |node|
+    @doc.root.each_recursive do |node|
 
       if node.text then
         a << [BacktrackXPath.new(node, ignore_id: true).to_xpath.split('/'),
@@ -30,8 +30,83 @@ class XMLColFinder
 
   end
 
+  def to_code(nametip: true)
+
+    @nametip = nametip
+    @tags = {}
+
+    xpath, remaining = @to_a
+
+    eid = getid(xpath)
+    linex = formatline('doc', eid, xpath)
+    a = scan(remaining, eid)
+
+    lines = a.flatten.compact.prepend linex
+    lines.join("\n").lines\
+        .map {|line|  line =~ /.text$/ ? 'puts ' + line : line }.join
+
+  end
+
+  def to_doc()
+    @doc
+  end
+
   private
 
+  def formatline(pid, eid=nil, key=nil, tail=nil, index: nil)
+
+    if eid then
+
+      nametip = @nametip && tail.is_a?(String)
+      klass = nametip ? key.scan(/@class=['"]([^'"]+)/).last : nil
+
+      line = if klass then
+        desc = klass[0][/^[^\-]+/].gsub(/(?=[A-Z])/,' ').downcase
+        desc += " (e.g. %s)" % [tail.length < 50 ? tail : tail[0..46] + '...']
+        "\n# " + desc + "\n"
+      elsif nametip
+        "\n# e.g. %s\n" % [tail.length < 50 ? tail : tail[0..46] + '...']
+      else
+        ''
+      end
+
+      key.gsub!("[@class='']",'') # Rexle XPath bug solution!
+      line += "%s = %s.element(\"%s\")" % [eid, pid, key]
+      if tail.is_a? String
+        line += '.text'
+        #line += "\n" if nametip
+      end
+
+    else
+      line = index ? ("%s[%d].text" % [pid, index]) : ("%s.text" % pid)
+    end
+
+    return line
+  end
+
+
+  def getid(rawtag)
+
+    rawtagx = rawtag.split('/').last[/\w+/]
+
+    tag = case rawtagx.to_sym
+    when :a
+      'link'
+    when :p
+      'para'
+    else
+      rawtagx
+    end
+
+    if @tags.include?(tag) then
+      @tags[tag] =~ /\d+$/ ? @tags[tag].succ! : @tags[tag] += '1'
+    else
+      @tags[tag] = tag
+    end
+
+  end
+
+
   # Groups xpath by matching branches
   #
   def group_by_xpath(a)
@@ -73,8 +148,6 @@ class XMLColFinder
 
         else
 
-          puts "path.join('/'): " + path.join('/').inspect
-          puts 'txt:' + txt.inspect
 
           h2[stickypath.sub(/^\//,'')] ||= []
           h2[stickypath.sub(/^\//,'')] << [path.join('/'), txt]
@@ -99,6 +172,64 @@ class XMLColFinder
 
   end
 
+  def scan(a, eid='doc', pid=eid.clone)
+
+    #puts 'a: ' + a.inspect if @debug
+
+    a.map do |row|
+
+      head, tail = row
+
+      if head.is_a? Array then
+
+        hline = scan(row, eid, pid)
+
+      elsif head
+
+        if head[0] == '/' then
+
+          key = head[1..-1]
+          puts 'key: ' + key.inspect if @debug
+
+          eid = getid(key)
+
+          hline = if tail.is_a? Array and tail.all? {|x| x.is_a? String } then
+            @prev_xpath = true
+            "%s = %s.xpath(\"%s\")" % [eid, pid, key]
+          else
+            @prev_xpath = false
+            formatline(pid, eid, key, tail)
+          end
+
+        else
+
+          hline = formatline(pid=eid)
+        end
+      end
+
+      if tail.is_a? Array then
+
+        if tail.compact[0].is_a? Array then
+
+          puts 'tail: ' + tail.inspect if @debug
+
+          tline = scan(tail, eid)
+
+        elsif tail.all? {|x| x.is_a? String} and tail[0][0] != '/'
+          puts '_tail: ' + tail.inspect if @debug
+          tline = tail.map.with_index do |x,i|
+            formatline(pid=eid, index: i)
+          end
+
+        end
+      end
+
+      [hline, tline]
+    end
+
+  end
+
+
   def truncate_xpath(records, offset=0)
 
     records.map do |record|

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: xml_col_finder
 version: !ruby/object:Gem::Version
-  version: 0.1.1
+  version: 0.2.3
 platform: ruby
 authors:
 - James Robertson
@@ -35,7 +35,7 @@ cert_chain:
   GSlZ9ilAfm8srTjbZ2cWQyNGGxH+zHQ3Z02c4ZEtgPv/wHjptd1VeBm0P1aemsRA
   ShsxXxzmzIrRENmpBp3tyR3k
   -----END CERTIFICATE-----
-date: 2022-01-23 00:00:00.000000000 Z
+date: 2022-01-24 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: rexle