🚧 WIP on replace-first algorithm - usr more non-capturing groups

Author: pmonks

src/lice_comb/impl/substitutions/bsd.clj

@@ -257,7 +257,7 @@
     #"(?<afterSystemics>Systemics(?<afterW3works>[\s\-–—]+W3Works)?)"
     #"(?<afterAMPAS>AMPAS|Academy[\s\-–—]+of[\s\-–—]+Motion[\s\-–—]+Picture[\s\-–—]+Arts[\s\-–—]+(?:and|&)[\s\-–—]+Sciences)"))
 
-(def re (re/join #"(?iuUx)(?<!\w)(The[\s\-–—]+)?"  ; Only public for ease of testing
+(def re (re/join #"(?iuUx)(?<!\w)(?:The[\s\-–—]+)?"  ; Only public for ease of testing
                  "\n\n#### Prefix ####\n"
                  (re/opt-grp (re-prefix-clauses) lcir/fre-mws)
                  "\n\n#### Leading clause ####\n"
@@ -266,20 +266,19 @@
                  (re/opt (re/alt-grp (re/ncg "beforeFreeBSD" "Free")
                                      (re/ncg "beforeNetBSD"  "Net")))
                  "BSD"
-                 (re/opt-grp #"[\s\-–—]*(style|like)")
+                 (re/opt-grp #"[\s\-–—]*(?:style|like)")
                  "\n\n#### Trailing clause ####\n"
                  (re/opt-grp lcir/fre-mws (re-bsd-any-clause "after"))
                  "\n\n#### Suffix ####\n"
                  (re/opt-grp lcir/fre-mws (re-suffix-clauses))
                  "\n\n#### Random dingleberries ####\n"
-                 (re/zom-grp lcir/fre-mws #"(variant|(Pub?lic[\s\-–—]+)?licen[cs]e)")
+                 (re/zom-grp lcir/fre-mws #"(?:variant|(?:Pub?lic[\s\-–—]+)?licen[cs]e)")
                  (re/opt-grp lcir/fre-version)
                  "\n\n#### Coda ####\n"
                  #"(?!\w)"))
 
 (def ^:private pairs-d (delay (concat [
   [re match->ei]])))
-;  (lcisu/spdx-match-pairs @ids-d))))  ;####TODO: confirm generic name/id matching is redundant for BSD
 
 (defn sub
   "Substitutes any BSD licenses found in the strings in `coll` with an

src/lice_comb/impl/substitutions/cc.clj

@@ -15,9 +15,12 @@
   without notice."
   (:require [clojure.string                     :as s]
             [clojure.set                        :as set]
+            [wreck.api                          :as re]
+            [spdx.licenses                      :as sl]
+            [spdx.expressions                   :as sexp]
             [lice-comb.impl.spdx                :as lcis]
+            [lice-comb.impl.regexes             :as lcir]
             [lice-comb.impl.utils               :as lciu]
-            [lice-comb.impl.parsing-utils       :as lcipu]
             [lice-comb.impl.substitutions.utils :as lcisu]))
 
 ;####TODO: IMPLEMENT ME!!!!

src/lice_comb/impl/substitutions/cpe.clj

@@ -22,7 +22,7 @@
 
 (def ^:private pairs-d (delay (concat
   (lcisu/spdx-match-pairs @ids-d)                         ; Generic license regexes handle most cases, except...
-  [[(re/join #"(?iuU)(GNU[\s\-–—]*)?(?:CPE|Classpath[\s\-–—]+exception)")  ; ...when no version is provided (and note that exceptions can't have "only", "+", "or later", etc.)
+  [[(re/join #"(?iuU)(?:GNU[\s\-–—]*)?(?:CPE|Classpath[\s\-–—]+exception)")  ; ...when no version is provided (and note that exceptions can't have "only", "+", "or later", etc.)
    (fn [m]
      {:id                      "Classpath-exception-2.0"
       :type                    :concluded

src/lice_comb/impl/substitutions/custom.clj

@@ -21,7 +21,7 @@
 
 (def ^:private pairs-d (delay [
   ; Proprietary / commercial
-  [#"(?i)(?<!\w)(Propriet[aoe]ry|Commercial|(Copyright\s+.{0,20})?All[\s\-–—]+Rights[\s\-–—]+Reserved|Private)([\s\-–—]+Licen?[cs]e)?[\s\-–—\.]*(?!\w)"  ; We consume - and . so that replacement doesn't leave them in and cause problems later on
+  [#"(?i)(?<!\w)(?:Propriet[aoe]ry|Commercial|(?:Copyright\s+.{0,20})?All[\s\-–—]+Rights[\s\-–—]+Reserved|Private)(?:[\s\-–—]+Licen?[cs]e)?[\s\-–—\.]*(?!\w)"  ; We consume - and . so that replacement doesn't leave them in and cause problems later on
    (fn [m]
      {:id         (lcis/proprietary-commercial)
       :type       :concluded

src/lice_comb/impl/substitutions/gnu.clj

@@ -106,7 +106,7 @@
                              [id confidence confidence-explanations]
                              [(str variant "-" default-version "-or-later")  ; Note: on the advice of SPDX technical team, default to "or later" variant if version not valid
                               :low
-                              (set/union #{:invalid-version-for-variant} confidence-explanations)])]
+                              (set/union #{:invalid-version} confidence-explanations)])]
     (merge {:id         (lcisu/assert-listed-id id)
             :type       :concluded
             :confidence confidence
@@ -161,7 +161,7 @@
                                       "\n\n#### Leading word salad ####\n"
                                       (re/zom-grp fre-gpl-words lcir/fre-mws)
                                       "\n\n#### Matching words ####\n"
-                                      (re/ncg "gpl" #"(?:GNU|GPL|(Genere?al(?:[\s\-–—]+Pub?lic)?([\s\-–—]+Licen[cs]e)?))")
+                                      (re/ncg "gpl" #"(?:GNU|GPL|(?:Genere?al(?:[\s\-–—]+Pub?lic)?(?:[\s\-–—]+Licen[cs]e)?))")
                                       "\n\n#### Trailing word salad ####\n"
                                       (re/zom-grp lcir/fre-mws fre-gpl-words)
                                       "\n\n#### Version and version qualifier ####\n"

src/lice_comb/impl/substitutions/hippocratic.clj

@@ -23,7 +23,7 @@
 
 (def ^:private pairs-d (delay (concat
   (lcisu/spdx-match-pairs @ids-d)                                                                     ; Generic license regexes handle most cases, except...
-  [[(re/join #"(?iuU)Hippocratic([\s\-–—]+Licen?[cs]e)?" (re/opt-grp lcir/fre-ows lcir/fre-only-or-later))  ; ...when no version is provided
+  [[(re/join #"(?iuU)Hippocratic(?:[\s\-–—]+Licen?[cs]e)?" (re/opt-grp lcir/fre-ows lcir/fre-only-or-later))  ; ...when no version is provided
    (fn [m]
      {:id                      (str "Hippocratic-2.1" (when (get m "orLater") "+"))
       :type                    :concluded

src/lice_comb/impl/substitutions/mpl.clj

@@ -23,7 +23,7 @@
 
 (def ^:private pairs-d (delay (concat
   (lcisu/spdx-match-pairs @ids-d)                                                                                       ; Generic license regexes handle most cases, except...
-  [[(re/join #"(?iuU)(?:MPL|Mozilla([\s\-–—]+Public)?([\s\-–—]+Licen?[cs]e)?)" (re/opt-grp lcir/fre-ows lcir/fre-only-or-later))  ; ...when no version is provided
+  [[(re/join #"(?iuU)(?:MPL|Mozilla(?:[\s\-–—]+Public)?(?:[\s\-–—]+Licen?[cs]e)?)" (re/opt-grp lcir/fre-ows lcir/fre-only-or-later))  ; ...when no version is provided
    (fn [m]
      {:id                      (str "MPL-2.0" (when (get m "orLater") "+"))
       :type                    :concluded