namespace DynamicBible.DataPreparation;

public class StepBibleProcessor
{
    public static void ProcessStep(string fileName)
    {
        using var fs = new FileStream(fileName, FileMode.Open);
        using var reader = new StreamReader(fs);

        Dictionary<string, List<Definition>> definitions = new();

        var start = false;
        while (reader.ReadLine() is { } line)
        {
            if (line.StartsWith("Reference + word	Word #	Greek", StringComparison.Ordinal))
            {
                start = true;
                continue;
            }

            if (!start || line.StartsWith("===", StringComparison.Ordinal) || line.StartsWith("#") || string.IsNullOrWhiteSpace(line))
            {
                continue;
            }

            // ok, at this point, you should be getting good data.
            var fields = line.Split('\t');
            var def = new Definition
            {
                Reference  = fields[0],
                WordType = fields[1],
                Greek = fields[2],
                EnglishTranslation = fields[3],
                Strongs = fields[4],
                Grammar =  fields[5],
                DictionaryForm =  fields[6],
                EnglishGloss =  fields[7],
                Editions =  fields[8],
                SpellingVariants =  fields[9],
                MeaningVariants =  fields[10],
                SpanishTranslation =  fields[11],
                SubMeaning =  fields[12],
                SuperMeaning =  fields[13],
                ConjoinWord =  fields[14],
            };

            if (!definitions.TryAdd(def.Reference, [def]))
            {
                definitions[def.Reference].Add(def);
            }
        }
    }
}

/*
   # Act.1.1	Τὸν	μὲν	πρῶτον	λόγον	ἐποιησάμην	περὶ	πάντων,	ὦ	Θεόφιλε,	ὧν	ἤρξατο
   #_Translation	The	indeed	first	account	I composed	concerning	all the things,	O	Theophilus,	of which	began
   #_Word=Grammar	G3588=T-ASM	G3303=PRT	G4413=A-ASM	G3056=N-ASM	G4160=V-AMI-1S	G4012=PREP	G3956=A-GPN	G5599=INJ	G2321=N-VSM-P	G3739=R-GPN	G0757=V-ADI-3S
   #_Non-NA variants
   #_Act.1.1 cont.	ὁ	Ἰησοῦς	ποιεῖν	τε	καὶ	διδάσκειν
   #_Translation	<the>	Jesus	to do	both	and	to teach,
   #_Word=Grammar	G3588=T-NSM	G2424=N-NSM-P	G4160=V-PAN	G5037=CONJ	G2532=CONJ	G1321=V-PAN
   #_Non-NA variants
   #_REFERENCE	WORD TYPE	GREEK	ENGLISH TRANSLATION	STRONGS	GRAMMAR	DICTIONARY FORM	GLOSS	EDITIONS	SPELLING VARIANTS	MEANING VARIANTS	SPANISH TRANSLATION	SUB-MEANING	SUPER-MEANING	CONJOIN WORD
   45_Act.001.001	=NA same TR ~~	Τὸν	The	G3588	T-ASM	ὁ	the/this/who	NA27+NA28+Tyn+SBL+WH+Treg+TR+Byz+NIV	τὸν		A la	the		»008:G3056
   45_Act.001.001	=NA same TR ~~	μὲν	indeed	G3303	PRT	μέν	on one hand	NA27+NA28+Tyn+SBL+WH+Treg+TR+Byz+NIV	μὲν		de hecho	on one hand
   45_Act.001.001	=NA same TR ~~	πρῶτον	first	G4413	A-ASM	πρῶτος	first	NA27+NA28+Tyn+SBL+WH+Treg+TR+Byz+NIV	πρῶτον		primera	first§1_first
   45_Act.001.001	=NA same TR ~~	λόγον	account	G3056	N-ASM	λόγος	word	NA27+NA28+Tyn+SBL+WH+Treg+TR+Byz+NIV	λόγον		palabra	word
   45_Act.001.001	=NA same TR ~~	ἐποιησάμην	I composed	G4160	V-AMI-1S	ποιέω	to do/make	NA27+NA28+Tyn+SBL+WH+Treg+TR+Byz+NIV	ἐποιησάμην		hice	to do§1_do/work
   45_Act.001.001	=NA same TR ~~	περὶ	concerning	G4012	PREP	περί	about	NA27+NA28+Tyn+SBL+WH+Treg+TR+Byz+NIV	περὶ		acerca de	about
   45_Act.001.001	=NA same TR ~~	πάντων,	all the things,	G3956	A-GPN	πᾶς	all	NA27+NA28+Tyn+SBL+WH+Treg+TR+Byz+NIV	πάντων		todas (cosas)	all
   45_Act.001.001	=NA same TR ~~	ὦ	O	G5599	INJ	ὦ	oh!	NA27+NA28+Tyn+SBL+WH+Treg+TR+Byz+NIV	ὦ		oh	oh!
   45_Act.001.001	=NA same TR ~~	Θεόφιλε,	Theophilus,	G2321	N-VSM-P	Θεόφιλος	Theophilus	NA27+NA28+Tyn+SBL+WH+Treg+TR+Byz+NIV	θεόφιλε		Teófilo	Theophilus§Theophilus@Luk.1.3
   45_Act.001.001	=NA same TR ~~	ὧν	of which	G3739	R-GPN	ὅς, ἥ	which	NA27+NA28+Tyn+SBL+WH+Treg+TR+Byz+NIV	ὧν		de cuales	which		»022:G0757
   45_Act.001.001	=NA same TR ~~	ἤρξατο	began	G0757	V-ADI-3S	ἄρχω	be first	NA27+NA28+Tyn+SBL+WH+Treg+TR+Byz+NIV	ἤρξατο		comenzó	be first
   45_Act.001.001	=NA same TR ~~	ὁ	<the>	G3588	T-NSM	ὁ	the/this/who	NA27+NA28+Tyn+SBL+TR+Byz+NIV	ὁ		el	the		»026:G2424
   45_Act.001.001	=NA same TR ~~	Ἰησοῦς	Jesus	G2424	N-NSM-P	Ἰησοῦς	Jesus/Joshua	NA27+NA28+Tyn+SBL+WH+Treg+TR+Byz+NIV	ἰησοῦς		Jesús	Jesus§Jesus@Mat.1.1
   45_Act.001.001	=NA same TR ~~	ποιεῖν	to do	G4160	V-PAN	ποιέω	to do/make	NA27+NA28+Tyn+SBL+WH+Treg+TR+Byz+NIV	ποιεῖν		estar haciendo	to do§1_do/work
   45_Act.001.001	=NA same TR ~~	τε	both	G5037	CONJ	τε	and/both	NA27+NA28+Tyn+SBL+WH+Treg+TR+Byz+NIV	τε		y	and
   45_Act.001.001	=NA same TR ~~	καὶ	and	G2532	CONJ	καί	and	NA27+NA28+Tyn+SBL+WH+Treg+TR+Byz+NIV	καὶ		también	and
   45_Act.001.001	=NA same TR ~~	διδάσκειν	to teach,	G1321	V-PAN	διδάσκω	to teach	NA27+NA28+Tyn+SBL+WH+Treg+TR+Byz+NIV	διδάσκειν		estar enseñando	to teach
 */

public class Definition
{
    /// <summary>
    /// Versification used by SBL and all modern texts.
    /// Significantly different from KJV at
    /// Mat.17.14; 20.4; Mrk.3.19; 12.15; Luk.1.74; 6.17; 7.19;
    /// Act.2.10; 3.19; 4.6; 5.40; 9.29; 13.39; 19.41; 24.18; Rom.7.9; 9.11;
    /// 2Co.1.6; 8.13; 10.5; 13.13; 13.14; Gal.2.20; Eph.1.11; 2.15; 5.13;
    /// Php.2.8; Col.1.21; 1Th.1.3; 2.6; 2.11; Heb.3.9; 7.21; 12.23; 1Pe.3.15;
    /// Rev.2.27; 13.1; 17.10;
    /// </summary>
    public string? Reference { get; init; }

    /// <summary>
    /// Source of the word, privileging NA, then TR.
    /// </summary>
    public string? WordType  { get; set; }

    /// <summary>
    /// spelling based on NA28 for NA, then TR if not in NA, then other Eds if in neither.  Cases and final accents based on
    /// the punctuation which follows THGHT
    /// </summary>
    public string? Greek { get; set; }

    /// <summary>
    /// English is based on Berean Study Bible, with permission, as at 1-July02019. This covered only NA words. Others supplied
    /// by Tyndale scholars, and the complete work was unified in many ways.
    /// </summary>
    public string? EnglishTranslation { get; set; }

    /// <summary>
    /// Extended with words not used in the KJV. Backwardly compatible with standard Strongs and NASB tagging, as defined in
    /// the Brief Lexicon at https://STEPBible\.github.io/STEPBible-Data/
    /// </summary>
    public string? Strongs { get; set; }

    /// <summary>
    /// Based on James Tuaber, with addtional details by Tyndale scholars: persons added to Personal, Reflexive and Possessive
    /// pronouns; distinguish 2nd verbal forms (e.g. 2nd Aorist); distinguished betweeen "Passive", "Either middle or Passive",
    /// "Deponent"
    /// - some changes have been made, including:
    /// ἐμοῦ  A-GSMN => P-1GS @ Jhn.016.015
    /// ευθυς =G2117 A-MNSM "straight" => G2112 ADV "immediately" @ Mat.003.016; Mat.013.20; Mat.013.21; Mrk.001.012;
    /// Mrk.001.028; Jhn.013.032; Jhn.019.034; Jhn.021.003
    /// V-PAI-3P changed to V-PAP-DPM except for  εισι[ν] (G1510) which is V-PAI-3P while οὖσι[ν] (G1510) is V-PAP-DPM
    /// τίς etc changed from I-NSM etc to X-MNSM etc at Act.025.014; 1Co.010.019; Heb.010.025, 27, 28; 11.040; 12.15x2, 16;
    /// 13.002; Jas.001.005, 7, 18, 23, 26; 2.014, 16, 18; 3.002; 5.013x2, 14x2; 5.019x2; 1Pe.002.019
    /// G4302 changed to G4277 when προειπον  is used, ie 45_Act.001.016-A, 49_Gal.005.021-A, 53_1Th.004.006-A
    /// G4302 changed to G4280 when προερέω is used, ie 41_Mat.024.025-A, 42_Mrk.013.023-INSTW, 42_Mrk.013.023-B,
    /// 46_Rom.009.029-A, 48_2Co.007.003-A, 48_2Co.013.002-A, 49_Gal.001.009-A, 59_Heb.004.007-IMNSTW, 59_Heb.010.015-B,
    /// 62_2Pe.003.002-A, 66_Jud.001.017-A,
    /// ποῦ =G4225 => G4226 at 42_Mrk.015.047-A, 44_Jhn.001.039-A, 63_1Jn.002.011-A
    /// </summary>
    public string? Grammar { get; set; }

    /// <summary>
    /// from the TBESG - Tyndale Brief lexicon of Extended Strongs for Greek at https://STEPBible\.github.io/STEPBible-Data/
    /// </summary>
    public string? DictionaryForm { get; set; }

    /// <summary>
    /// from the TBESG - Tyndale Brief lexicon of Extended Strongs for Greek at https://STEPBible\.github.io/STEPBible-Data/
    /// </summary>
    public string? EnglishGloss { get; set; }

    /// <summary>
    /// Those which use the same letters, though they may be accented or capitalised differently. Byz=Byzantine based on RP;
    /// NA27=Nestle-Aland 27th ed; NA28 2012 (this is NOT the same as ECM Acts - see eg Act.1.10 ἐσθῆτι in ECM); TR= Textus
    /// Receptus; SBL= SBLGNT; Treg= Tregelles; WH= WH;  Tyn= Tyndale House GNT.
    /// </summary>
    public string? Editions { get; set; }

    /// <summary>
    /// </summary>
    public string? SpellingVariants { get; set; }

    /// <summary>
    /// </summary>
    public string? MeaningVariants { get; set; }

    /// <summary>
    /// Spanish is based on  Marvel Bible Project as on 9-Jan-2019 from
    /// https://github.com/eliranwong/OpenGNT/blob/master/OpenGNT_BASE_TEXT.zip &  OpenGNT_keyedFeatures.csv.zip. This was
    /// available only for words in NA28. Other words are supplied by Tyndale scholars.
    /// </summary>
    public string? SpanishTranslation { get; set; }

    /// <summary>
    /// Words with more than one meaning are supplied with a context-sensitive sub-meaning. Individuals and Places are
    /// identified as in TIPNR. Other are based on CSG (Context Sensitive Gloss) from Marvel Bible Project as at 9-Jan-2019
    /// from https://github.com/eliranwong/OpenGNT/blob/master/OpenGNT_BASE_TEXT.zip &  OpenGNT_keyedFeatures.csv.zip. This was
    /// available only for words in NA28. Other words are supplied by Tyndale House who also edited the CSG for words where
    /// sub-meanings  were not sufficiently detailed.
    /// </summary>
    public string? SubMeaning { get; set; }

    /// <summary>
    /// </summary>
    public string? SuperMeaning { get; set; }

    /// <summary>
    /// Links words that might not be translated (eg articles and particles) with the word they are connected to, which are
    /// often separated by several other words.
    /// </summary>
    public string? ConjoinWord { get; set; }
}