jst-parser 0.1.0 → 0.1.1

checksums.yaml

@@ -1,15 +1,15 @@
 ---
 !binary "U0hBMQ==":
   metadata.gz: !binary |-
-    YWM5MzYzYmZkMmU0ZDY1YzZkMmY4NmE0YmM2MjE3MTlhYWQ2MzYzOA==
+    YmZlMDNmYjY5MTYzYTFjNmQ5NjFiMjhkOTY4NDkwYjhjMWZiOTg3OQ==
   data.tar.gz: !binary |-
-    ZjE0Yjc2YTNmMGQxZjNiNGYzOGQxMzk5OWY1ZDZjZmE4NGY1NjQ2Mw==
+    ZjI2MGI5ZjlmMGFiYWI1MWUzOWFlOTI4MzAwYTk5YzBjY2M0YjY3ZQ==
 !binary "U0hBNTEy":
   metadata.gz: !binary |-
-    MWNjMGQzYzQ4ZmZmYjcyNjM5ZTJkMzdiMmEzMGZmNGJmMzUwZWM3NzM0MWMx
-    YTMzZWUyNmI5ODI5NzU1ZjM3MmMxNGNhNjU2YjgyYTg2NWM3OWU5MjI1NTVm
-    Njc5MWM2YWFmYjA5ZjhkNjdkZDYzNzJjNjZlNTI3NGFkOTYwYmY=
+    MWNhYWQ2ODA4Y2FiOWQxOTNiNjUzMWNhOWRiYzk0NmM3MDVhODdlMDE3YmRk
+    MzE5MWVmZDU3YmE0OGJkYWU1NTZjNWZhZGRmMGJjYTI3MzUxMDhhMGExZWFj
+    ZDBhODVmNzJiZGQyODhhYjFjOGYwNGE5MjVjMmUwZjFkMDRkYTY=
   data.tar.gz: !binary |-
-    NDUyNDNmMDVlOTBjYmYyYWMzZWNmMDg2NDgwMzIyMjgzYWVjMjk1ZWQ0MDVk
-    MjBjY2UwM2NiZTZhOTc2MTEwMzQ3M2IyMWVjODRmNTk0MDhkYWY3ODJmNjAy
-    NzgwZjI3NjA1MjI1MzgyOTdhZTg1MzkxYTAzMzdjNzQzOTEwNzk=
+    ZjA0YzY4MDJlNDY5ZGU1N2FjZjYyMDFjY2U0NzQ2YjgyNGY0YTRjZDk5ODQ3
+    YjM2OWE3YTA4N2ZjMjNiY2YzODg3MjM4ZTE0MzZjYjA5ODFhMTE2ZGE3ZGM3
+    ZmE4NmY2MzA2ZDM4YzcyNjg0YzY2OWY2YzM5YThkZDkzOGI4ZDI=

data/lib/jst.rb

@@ -14,6 +14,7 @@ module JST
 		BRANCH_DOD = 'Department of Defense'
 
 		class BadPDFError < StandardError ; end
+		class UnknownPDFParsingError < StandardError ; end
 
 	 	def parse(pdf_file)
 	 		unless @debug
@@ -48,10 +49,12 @@ module JST
 					raise BadPDFError, "Could not parse JST."
 				rescue ArgumentError
 					raise BadPDFError, "PDF text parsing exception."
-				rescue PDF::Reader::UnknownGlyphWidthError
-					# Waiting for this exception to be commited from the following pull request:
-					# https://github.com/yob/pdf-reader/pull/105
-					raise BadPDFError, "PDF text parsing exception."
+				rescue => exc
+					raise UnknownPDFParsingError, "#{exc}"
+				# rescue PDF::Reader::UnknownGlyphWidthError
+				# 	# Waiting for this exception to be commited from the following pull request:
+				# 	# https://github.com/yob/pdf-reader/pull/105
+				# 	raise UnknownPDFParsingError, "PDF text parsing exception."
 				end
 			end
 	  	end
@@ -68,10 +71,12 @@ module JST
 			skills_upper_regexp = /(.+)\s+(\d)\s+\w{2}\s+U/
 			skills_vocational_regexp = /(.+)\s+(\d)\s+\w{2}\s+V/
 			skills_graduate_regexp = /(.+)\s+(\d)\s+\w{2}\s+G/
-			ignore_privacy_regexp = /PRIVACY ACT INFORMATION/
-			ignore_date_regexp = /\(\d{1,2}\/\d{1,2}\)\(\d{1,2}\/\d{1,2}\)/
-			ignore_misc_regexp = /None|NONE ASSIGNED/
-			ignore_orphaned_skills = /^(\d|L|U|V|G|SH)$/
+			ignore_regexp = []
+			ignore_regexp.push (/PRIVACY ACT INFORMATION/)
+			ignore_regexp.push (/\(\d{1,2}\/\d{1,2}\)\(\d{1,2}\/\d{1,2}\)/)
+			ignore_regexp.push (/None|NONE ASSIGNED/)
+			ignore_regexp.push (/^(\d|L|U|V|G|SH)$/)
+			ignore_regexp.push (/\*\*/)
 			ignore_list = nil
 			@positions = []
 			@skills_all = {}
@@ -96,15 +101,15 @@ module JST
 				if line.match(experience_section_start)
 					# Reached the job experience section.  Begin parsing out.
 					inside_experience_section = true
-					puts "-- -- -- -- -- -- --  INSIDE EXPERIENCE, PARSING -- -- -- -- -- -- -- " if @debug
+					puts "-- -- -- JOB EXPERIENCE SECTION START -- -- -- " if @debug
 					next
 				end
 				if line.match(experience_section_end)
-					puts "-- -- -- -- -- -- --  FINISHED PARSING -- -- -- -- -- -- -- " if @debug
+					puts "-- -- -- JOB EXPERIENCE SECTION END -- -- -- " if @debug
 
 					# Finished last job position.  Appent previous job position.
 					if !position_title.empty? && !position_desc.empty?
-						puts '-- -- -- -- -- -- --  APPENDING POSITION -- -- -- -- -- -- -- ' if @debug
+						puts '-- -- APPENDING PREVIOUS POSITION -- -- ' if @debug
 
 						append_position(position_branch, position_date_begin, position_date_end, position_title, position_desc)
 						position_branch = ''
@@ -117,8 +122,9 @@ module JST
 				end
 
 				if inside_experience_section
+					line.strip!
 					if line.match(experience_regexp)
-						puts "~~~~~ NEW EXPERIENCE: #{line}" if @debug
+						puts "-- -- NEW EXPERIENCE" if @debug
 
 						# Determine which branch this job title falls under
 						position_branch = BRANCH_ARMY if line.match(/AR-/)
@@ -130,9 +136,10 @@ module JST
 
 						# Determine the service date (dd-MMM-yyyy)
 						if date_match = line.match(experience_date)
-							puts " ^^^^^ PARSING DATE ^^^^^^ "
 							position_date_begin = date_match[1] unless date_match[1].nil?
+							puts "-- START DATE:  #{position_date_begin}" unless date_match[1].nil?
 							position_date_end = date_match[2] unless date_match[2].nil?
+							puts "-- END DATE: #{position_date_end}" unless date_match[2].nil?
 						end
 
 						# Next line will be the job titles
@@ -152,12 +159,12 @@ module JST
 					end
 
 					if at_job_title
-						puts "~~~~~ JOB TITLE: #{line}" if @debug
+						puts "-- JOB TITLE: #{line}" if @debug
 						at_job_title = false
 						position_title = line
 
 						# Next line will be the job description starting point
-						puts "~~~~~ AT JOB DESC" if @debug
+						puts "-- JOB DESCRIPTION:"
 						at_job_desc = true
 						next
 					end
@@ -177,7 +184,7 @@ module JST
 							# Strip out skill name
 							skill_name = skills_match[1].strip!
 
-							puts "**** SKILL: #{skill_name}" if @debug
+							puts "-- SKILL: #{skill_name}" if @debug
 
 							# Init skill name key, if none exists
 							@skills_all[skill_name] = 0 if @skills_all[skill_name].nil?
@@ -200,9 +207,17 @@ module JST
 					end
 
 					if at_job_desc
-						unless line.match(ignore_privacy_regexp) || line.match(ignore_date_regexp) || line.match(ignore_misc_regexp) || line.match(ignore_orphaned_skills)
-							puts "m: #{line}" if @debug
-							position_desc += line
+						ignore_line = false
+						ignore_regexp.each do |regex|
+							if line.match(regex)
+								ignore_line = true
+								break
+							end
+						end
+
+						unless ignore_line
+							puts "-> #{line}" if @debug
+							position_desc += " #{line}"
 						end
 					end
 				end
@@ -216,12 +231,8 @@ module JST
 			position[:date_begin] = date_begin
 			position[:date_end] = date_end
 			position[:title] = title
-			position[:description] = description
+			position[:description] = description.gsub!(/  /," ")
 			@positions.push(position)
-
-			# @positions[position_title] = position_desc
-			# position_title = ''
-			# position_desc = ''
 		end
 
 		def create_response

data/lib/jst/version.rb

@@ -1,3 +1,3 @@
 module JST
-    VERSION = "0.1.0"
+    VERSION = "0.1.1"
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: jst-parser
 version: !ruby/object:Gem::Version
-  version: 0.1.0
+  version: 0.1.1
 platform: ruby
 authors:
 - Chris Little