From 6d689d8760bf32c53c30c22e2a48a1740c0b4e04 Mon Sep 17 00:00:00 2001
From: roblabla <unfiltered@roblab.la>
Date: Fri, 6 Jan 2023 11:26:15 +0100
Subject: [PATCH 1/2] Add COLNAMES option to give names to the columns

The COLNAMES option should contain the row number giving the name to the
columns. This name will be used in place of the lettered column name
used by excel in the SQL Table.
---
 src/options.rs             | 14 +++++++++++++-
 src/spreadsheet/manager.rs | 24 +++++++++++++++++++++++-
 2 files changed, 36 insertions(+), 2 deletions(-)

diff --git a/src/options.rs b/src/options.rs
index bfccb19..12fcb01 100644
--- a/src/options.rs
+++ b/src/options.rs
@@ -1,6 +1,6 @@
 use nom::branch::alt;
 use nom::bytes::complete::{escaped, tag, tag_no_case};
-use nom::character::complete::{alpha1, digit0, multispace0, multispace1, none_of};
+use nom::character::complete::{alpha1, digit0, digit1, multispace0, multispace1, none_of};
 use nom::combinator::{map, recognize};
 use nom::{IResult, Parser};
 use nom::sequence::{delimited, preceded, separated_pair, terminated, tuple};
@@ -9,6 +9,7 @@ pub enum UsingOption {
     File(String),
     Worksheet(String),
     Range(String),
+    ColNames(String),
 }
 
 pub fn parse_option(input: &str) -> IResult<&str, UsingOption> {
@@ -16,6 +17,7 @@ pub fn parse_option(input: &str) -> IResult<&str, UsingOption> {
         parse_filename_option,
         parse_worksheet_option,
         parse_range_option,
+        parse_colnames_option,
         ))).parse(input)
 }
 
@@ -49,6 +51,16 @@ fn parse_range_option(input: &str) -> IResult<&str, UsingOption> {
         |t: (&str, &str)| UsingOption::Range(t.1.to_string()))(input)
 }
 
+fn parse_colnames_option(input: &str) -> IResult<&str, UsingOption> {
+    let option = tag_no_case("COLNAMES");
+
+    let value = preceded(
+        tag("'"), terminated(digit1, tag("'")));
+
+    map(separated_pair(option, multispace1, value),
+        |t: (&str, &str)| UsingOption::ColNames(t.1.to_string()))(input)
+}
+
 fn parse_with_spaces<'a, T>(parser: impl Parser<&'a str, T, nom::error::Error<&'a str>>)
     -> impl Parser<&'a str, T, nom::error::Error<&'a str>> {
     preceded(multispace0, terminated(parser, multispace0))
diff --git a/src/spreadsheet/manager.rs b/src/spreadsheet/manager.rs
index 15a6c76..ce19387 100644
--- a/src/spreadsheet/manager.rs
+++ b/src/spreadsheet/manager.rs
@@ -7,11 +7,13 @@ use calamine::{open_workbook_auto, DataType, Range, Reader, Sheets};
 use std::fs::File;
 use std::io::BufReader;
 use std::path::Path;
+use std::str::FromStr;
 
 pub struct DataManager {
     sheets: Sheets<BufReader<File>>,
     worksheet: String,
     range: Option<CellRange>,
+    colnames_row: Option<u32>,
 }
 
 pub enum DataManagerError {
@@ -49,9 +51,17 @@ impl DataManager {
     pub fn get_columns(&mut self) -> Vec<String> {
         let range = self.get_effective_range();
         if range.get_size().1 > 0 {
+            let row_workspace_sheet = self.colnames_row
+                .and_then(|v| Some((v, self.sheets.worksheet_range(self.worksheet.as_str()))))
+                .and_then(|(row, sheet)| Some((row, sheet?.ok()?)));
             (range.start().unwrap().1..=range.end().unwrap().1)
                 .into_iter()
-                .map(|n| CellIndex::new(n + 1, 1).get_x_as_string())
+                .map(|n| {
+                    row_workspace_sheet
+                        .as_ref()
+                        .and_then(|(row, sheet)| sheet.get_value((*row, n)).map(|v| v.to_string()))
+                        .unwrap_or_else(|| CellIndex::new(n + 1, 1).get_x_as_string())
+                })
                 .collect()
         } else {
             Vec::new()
@@ -70,6 +80,7 @@ pub struct DataManagerBuilder {
     file: Option<String>,
     worksheet: Option<String>,
     range: Option<CellRange>,
+    colnames_row: Option<u32>,
 }
 
 impl DataManagerBuilder {
@@ -91,6 +102,11 @@ impl DataManagerBuilder {
                 UsingOption::Range(range) => {
                     builder = builder.range(CellRange::try_parse(range.as_str()).unwrap());
                 }
+                UsingOption::ColNames(colnames) => {
+                    // We substract 1 to go from excel indexing (which starts at 1) to 0-based
+                    // indexing of the row.
+                    builder = builder.colnames_row(u32::from_str(colnames.as_str()).unwrap().saturating_sub(1));
+                },
             }
         }
 
@@ -112,6 +128,11 @@ impl DataManagerBuilder {
         self
     }
 
+    pub fn colnames_row(mut self, row: u32) -> Self {
+        self.colnames_row = Some(row);
+        self
+    }
+
     pub fn open(self) -> Result<DataManager, DataManagerError> {
         if let Some(file) = self.file {
             if let Some(worksheet) = self.worksheet {
@@ -120,6 +141,7 @@ impl DataManagerBuilder {
                         sheets,
                         worksheet,
                         range: self.range,
+                        colnames_row: self.colnames_row,
                     }),
                     Err(err) => Err(DataManagerError::Calamine(err)),
                 }

From f73381f43400ef90c73b60a0f63aecbdeee6c025 Mon Sep 17 00:00:00 2001
From: roblabla <unfiltered@roblab.la>
Date: Sun, 8 Jan 2023 15:37:38 +0100
Subject: [PATCH 2/2] Add tests for colnames feature

---
 src/options.rs             |  13 +++++++++++++
 tests/abcdef_colnames.xlsx | Bin 0 -> 5372 bytes
 tests/lib.rs               |  37 +++++++++++++++++++++++++++++++++++++
 3 files changed, 50 insertions(+)
 create mode 100644 tests/abcdef_colnames.xlsx

diff --git a/src/options.rs b/src/options.rs
index 12fcb01..ae34a6d 100644
--- a/src/options.rs
+++ b/src/options.rs
@@ -132,4 +132,17 @@ mod tests {
             _ => panic!("Expected range option")
         }
     }
+
+    #[test]
+    fn parse_colnames_option_produces_colname() {
+        let (output, option) = parse_colnames_option("COLNAMES '152'").unwrap();
+
+        assert_eq!(output, "");
+        match option {
+            UsingOption::ColNames(colname) => {
+                assert_eq!(colname, "152");
+            },
+            _ => panic!("Expected colnames option")
+        }
+    }
 }
diff --git a/tests/abcdef_colnames.xlsx b/tests/abcdef_colnames.xlsx
new file mode 100644
index 0000000000000000000000000000000000000000..477c08c2693d8dec308972c79a00f673430f288c
GIT binary patch
literal 5372
zcmaJ_1z1$;)}9fN9CB!o?i$jeQAE1C1c3n&h9O2eq`N~H5s(y>kWfjL5|k9B89-7b
zBxI!T;5q-1>-~>wKYQxg>zi+{wf6hIYw2m?;DP}J1O$K<FJ2SCIU~k?w(*3ydWi_1
zy;h{X(P@W}L~RDd@na;Xl?Wb|HfgCfu}NHd&FRKiA6wKBzP)#iijpL;raj3cION2x
z1g+&Jh=y=$-iS3ekG;^sZ68$D@&3faPeiJ(Fu02Xs%n5x=J!n2)!6NwyYj9$0GXme
zw6deEIRsp`UvmJ+D@A)+8~cczIK(VPn6D*JdN`RBj)mYhM`*)cB;C|mnEVyQKbx|r
zvw3%GXcYyPL7OQ-uqYxwG%Tp${mf<tDLn~9O`GB#+tNli&G3sPw&RTVU$V2Q8XJ~8
z-R*WS_BB&RK%9^P+8jY@S{wP)%;Gj0@fLYU@0Y0bH1P={Q4**PY!Lb{004Ub8YVLA
z6~3+_J_t`2I|RZ-*ca}aXgKLP0U;0CH1*vjZFT2^WRpnxrsbfyi+gHatDeOP`W162
z$sA)UVY#Kl3cJAKmX)>f>G{RJhAgx=oj_fjIg}(DpMNw`?=U+kKn-7f$$a7Qtbstx
zIG$jWd5dLUowS1mi=2OzYcGF?<J_TCRl$<?BSt+iUwyfBn)6;Ooa9@Z3BXKKktnR(
z#_OS3bT#BO{Na0ph3cXGXBQ)xV}L!WW0M*ZSru6&Dc{B<7q&|GdtC}2xaWw6eAhU9
zvL%+slBBIME-KJar<7=Nu*-EwDs7-@rLjcK#FmRcudql)keHzk>mHZ6gqORv>FXlX
zVF~YuH|bs!zHIR>`=RuWClvbN{a5Tw-g;zmwJkeGoJJ3IMT@Ad(Nnq`2b@GtZ-ovI
zxlSFrU0D-I7#GY{?N!d3ei0rw&RO780->>{h;A!ScPj9&Pyy{DWIScR>gqOxHr~6b
znc3eI#xMJTH8L-fI)}5%k&V}OBd(im5AxAXXNQ&`YMUy^=o*`=G(E54moV0m2=ByK
z!ppZ2O6)vo?W_cCG_?Ich6-h?4AM{=AwBAGy-GZ+Wn4?li815JnBiNf-qImfi|C>&
zbZ$K1WJOyJDQ?3V-1l75{0`EQi+Z%XnldLepde0+-iC19q*vLqJ1^yzSx1v{bbF>>
zajan8nH^s^D;+cNR{aSO0Px27Q|W;IDjhGRpDWDktZ>%NtP%Gi;GmhR>Z8`HA0(`K
z3RY!A*I|!Gj8GEmZ}4zTjze9`K*!&kx+0Vmp9>Gy8EM>bW!aF%xWlQGk&Tp0(+2SD
zG|@b!2s-_o^4uP=F4Oqkb=iajE-@Y}b%yyUhCozvYyuqG@|3rp09+jF0fQea5NMg`
zU+E2fTpl41`60}VZdD9YU&?toiPIMXYWHI$NQcHQ7!<cLWR#cG*%n2oAQYAX{Z4d#
z4PKeynYu;>+ECIo-WA-mM|z$`UJ>#s*Oo3#Q<AbjBpv{pGJM&V>2hXxu86P6{)kL=
zxR;pr8U*b-Xedw!nyNE(@j4{YUG$5c75>0;`dAG<?-=WW9{2zbV2!<PC-g}y61aL0
zUm8`ORlV1&#fZ7TVvz1ACeJ`VadG>Gz=L)RmP&E5hq<W+>TOl4bhHtSaRDnujp;6s
zSg4R7fs2VrW&LB)+?aX|C8s`*@)pmT&|Uh)qYv<0MCa0mGPyN%GHs(Pel^PjOy$?i
zMqs%My7*nJSxYhVw}&A5D^9tY27~yav=-*5aqpQI?v$x_SRZFNVp!L?tJn5&0<_Gs
zq`PNCu4h%QnZjaCfx14ZkHmmOcCnlnv2&<$iS*0FnKdV38RZ*r*2)@2RF$@gApR<=
zm>D%0KegMc#Q1Za2LsWO3~lWi75CnJ6&0Vp@7L8k65xVx{J(=yytdu?pe$G0`(2-F
z*_4(~PhU@n6ctJG<_Sbm?kyr9!G=2d0on0U$@fw3QkG2ko)au!KjPH^rl7zwIQ-tb
zV7~q>fXYUse3#8sG?fDH<o7g>_r)xD$%M-q1T{(dOo;^11C)c9Z>qYeWVIb0b<K+M
zUR0i!+~?llPWnav%b%GQJDUn*!lhUEcsSxtt*ZIBNu*|8rsbd2d2UzYdIc69d8z(X
z=L~<<`Cnvm9w`4JmQ-z*pTu%ZeI3)Oj&Gj;ZArLNb^HU^rT=-bLn>{g{xn#a(UJ(Z
zHkdv*y3?xmeN`fHHKs$E<l^H(+{ibb-UKb7#~bT&^d^0U=-FFb$<-{BPnxO5?$WA@
zSk&%7&4umSQe>@*do*;)M|&E>*Y!o-Z>taI0{f^9D=cy9+ywNRyxLBy?MFeRHbnvX
zkZurRpXa4i@gaj`yIBDdG5lTl{-XS%B0KXK$GfWVd;=*T!cR7imp1*05rmT!dckW)
zoIBeua#Sj+(9?lG5kU9bzW)d8eRKVhJm`Lrhu19_4Cy8E^G);&3|dU#$)BajPtZ0L
zi-)PStbziwNc<90=c?RyH*{Aor$aj|-s-?{nhw9pdfy<B3a~Jy4%Be$8t9c1C)9Oh
z$=6(EewGqZ+rR?4I@2z=NPqLN;q=(!c&;B)d{t06LtGAHQ@yJw9GvsurEeNu*E<{l
z_j2$<qdP3PVnXh*3wKSkLwS}9uMxR*Dx_q-<`s|DVCcA+jmyQ}TSc+a836IkO9xl^
z`tyn!O2)hf^YPly6sHMIc<l){YZXw{O^gf+co+xBG~Zg7&E%xE+O`52+>k11u+|Cx
zqLM$reu7XNqO*{vlrMjYCoqgbnzL^qA1MI5@wC<nxK=wQ)<hgHp5FiFd6k;R#_ytX
zKSd^8Aoaw&kjad|ylL_gBb|j$bv9n#T&Exhgv>gJ$6W!VsDgPCn*;%dPz0v6xaiUx
z)kPhUf_rHC=|ZSD16WDIjboMVkJ9tAa)W`ww%UhDuS8Qu7;wjF_2zK;jq%F|-QxLl
zsQV9=rSMhQdiF}DJSKn+P*wcLNWzq)_>tg<xMhW5q`}>VOr2*P8LLu3BA=eU!U-1J
zZZYkD50B$b0Y~24JR$R%dVX7mWsq2r&Vr#TxW+qI(K)>f^gTt37KJAIQdV8wh9Ngg
zep$}FF*B+vzppEw@<UDpH1*0&wN{D|r^qcEKogvzMXhHlrxvwk4NjIt+49;{mohca
zzrkgyf+UW7{5IfH$nuTC5Nb=bI?1X$ZR4)5WH>;?SC>r~ylqQ^`Vwx`{4~4~TzYp7
zfHuJE5$lyvWj;<1(~^Y9$MKhAimVTe%7)W3xdoWAY~_X9%-th2hPgl2GE!Jy7o3jy
zA!fIDounxSAhS7<95NEzcX~P*vSuVzbFise-53&ISVAoJoWPwU?JF(`Li_b#xvL!o
z1W8zZ`88fcE%x`)Q9`<D)ZNsknLQ4OCw;Y4pdVf*yYpgH^D~@zOGAiv{ELG_Id48X
z3$WFdXID+!x*07~rchxWLg-)$PGLNg?royDV{Uh4^jKI=1z+~dP97ROs(Tlnyjr9N
ze~qm1R9+Jz@DEAIWYvQ67X`Ll_(<{SQ!Q#DKQjXkR2jW}Ka6<$UCvb`p(jJU*UrIx
zrPPmiaO(w+;yUt_>~8OxK(iCLhk2bt(yu35f4}QHwD)D2H-TrJ?BunN)tl_fMk0=r
zIu|AtAAN9FJ7-cmMZse3ldqKpRMpW$NW;j^tmZIZ)P+9zM8^oY01`E+Gw~xYYK$OV
zex7H~%=S6Z%{CMlc)R!B+o`ITdKW149Ax7@l<5i{>XUzRhLN?3`!`D91wCIVIf+QM
zu&9aGf69O1)Tgn|`=XQPJAtsiG$kKN0T<=F_fqAerA6zxujU!vS<=or_Loxx$I}ng
zSQxfje6O6Vx&E|o`i>&~l=n<Xd<gIij9X*5p+dL-fHLWyz!?0u=(u&$6Xsxu^mKM}
z`l&q<)Bo1@aw-e5nxi<VQVA;8HNsio)vuKlvBRCt_z2y*RJ034>jU}USzJk3L?2O$
zA6{d2y$s?lRsiJ*GdH*9Zu%6wgfdvEMKA}lx4%KtFBPuxd*B=L?c8wL;7Lfj?0o%k
zy6EDHhR}}6PVNv$jxCE0IVnLVBcoezP^b5$0DeVip^ErY2;(HPSvwk)&k>MkZG$i+
zw!cB~lz3+$L3m^ZvC5K`^&`qW0&Q^V+H4aK{SV$kqO_QHV*-U9<P|}!aKTOeEKJl*
z6$<|}LY{nLqFTB5N1{TWi{_=f^EO45iVm-^Y961U;FkTxHEb+G|0vNYuva)B?Daem
z?p`AH2v67<GQ$me+`=K`Ef|56O<PutNOJP>3eVfc8tNY(UmndPe=%mZ=Wuw)pmwWp
zNeZ#WI55_^_6TK5Hqv8qk2BVe^6pYaN~3U(za*JBCDg&9y7ZG?8WmLR$t5B;Pqc+^
zZ3Qr%(rPxm?tKGa56#d*hD<A7r_}=UMEv~`{-{e6D2A>OGmu2$yy{~>j&1iXS~`?y
zYAy#J3rAnSRM6d7FoRyf&=9A_N<Qryejkidm6dd6M|x<}NIbMLStqHheU6-&yXDi2
z;zO4BQJCUw-a9Y&NsIY$2O99ADk_NlL4iWAJp@1pnPi_qa~|!;mYu_LrRc9P1bw;N
z4#?}}H$+`3Yo3n6JgE=*T4lf0tZH|3sK|%Ig?*SLk1<18Iz*=pYRNS$*S8HlTtlHR
z?miG&gJFyxJI^2^rYoj5zA{2dITBxw@;(a<ZOu4s)%{NVQ%fLqCJnHfIqr{)Nq#;d
zH{IRO^6?<mQnw8Pemipv)|?k`@o7&)a$U0z9JF<(oKbTgC~$PSUr+gCk0$1I?edqY
z?ZbTH2OEdN6K`fSh9ouY!q^Af)JiT(9=PVIa>eB?TWt5=dNc%NLe<C~D9{myo6Z(0
z4OU*a++4@C=XpNpMH^j1nmIxuk6~ky8mp_I{N~W1OExi|ks=z^f-iI>X$>!?zoV!B
zl$Jjc+P|FD_vYfIBGB&h`Ac-2rY}VEd=ghfOW1-4S9&LO2sheQ#-3P^K@`<zqoK*Q
z(3X&SyOXVYR3A;U$h=&mvw0*}>lNSJl;#DP35>}AKyJ)8DWy3QsaB%}hO?_T2n$@@
zsu<Jv^9#n)jLXKAl{{MHZN2f}uO_61YfYRA8_-7V`S1NJR=J(4@H<{e1pH@A`=1eE
zV;VU(T}CiIX@sBpp)MXOYlKNbKL>gty0znj0yx4plA>7@bcW@Rm7kK2Q`~M45#R#|
zfk=6T5XU7@^*24UwwEFb7B31md5>Cg;KlR+;{{n=q|3m!OUC+VLHDHXg*j9Z26-L^
z5=s4SIcuMR=`u#2^v5Hq_Spla_qSFhk2OghW;TDI5o`hQqF|><+OtBL8v)$Chu!rS
z_WbjGSt=vkkT5r-jgg-_%**<xs>*ne9b=HIV2<OuEx}E&P;!k)ykst;!l$Vo+_BC_
zDeGB#oAf+-#^eUZiDu?|y)90Tg0lA!#t&Ry`#JAsXsu*(7DV=x2$wCu_ET=8@4IFQ
z`OpN0HB9KLQ-`(O?2RV^V?LS<((xr2%FT{s_If`1%vJq<>qi`65zb0V)XEjdab|F!
zb@W1T{97T~Ma~BJ4*PA5m-Y2vIos?`-mf2}9|a}57wxcoDzVAbH<FTWO&hT}(ynei
z0;ceL(+o6GZ_+e!X#*s6p>8~0YBW=ks+Ew?mM$GS=9l@xAD%n1bXA}e%!f{%`3!35
zs{tD*Tp}q~D{%4-iC7<@qy&tuRyhJ$yd{4;{F3;E!c%ss{p8eRLAxYR^1Z|C2Il1L
zbsG=-eJ}SPt@y7(l{Z^u^j^7l`5)0U<=*JP92Lpb-@+B{K6x@jHFU66$=3RvY1?qz
zZarMzY2Nbw%s2fml#Tx$;aQsAR^I>Sh%LR3cxPz>UH}7r&6&=RR?g;3|83`{PQN>z
zZ>-NIFTacj`&YlsVSaZ$->sbW-oMNaO9EKuU!C~xuIJ_aSv&K~G%o&h{g>wE_jS&5
z^qKhoWe)iN?EimSf!|j-&s}GH{>w(O(LB46-`V~56>`sk?HR}YGCXW#{)h2?cRvp;
zXNdgEvasUgANRlU`FHR05AYfM{IX?iPXF<?{|AhIU*LRc{r3W|u)Fzxo_ak^JZzc(
Q03z&h7t1e>lxN@m1J(A~8vp<R

literal 0
HcmV?d00001

diff --git a/tests/lib.rs b/tests/lib.rs
index c339b38..7149050 100644
--- a/tests/lib.rs
+++ b/tests/lib.rs
@@ -141,3 +141,40 @@ fn test_abcdef_file_aggregate() {
         AbcdAgg { kind: "odd".to_string(), count: 3 },
     ]);
 }
+
+#[test]
+fn test_abcdef_file_with_colnames() {
+    let connection = init_connection();
+    connection.execute("\
+        CREATE VIRTUAL TABLE test_data USING xlite(\
+            FILENAME './tests/abcdef_colnames.xlsx',\
+            WORKSHEET 'Sheet1',\
+            RANGE 'A2:D7',
+            COLNAMES '1'
+        );\
+    ", params![]).unwrap();
+
+    let mut query = connection.prepare("\
+        SELECT alpha, number, word, kind FROM test_data;\
+    ").unwrap();
+
+    let rows = query.query_map(params![], |row| Ok(Abcd {
+        alpha: row.get(0).unwrap(),
+        number: row.get(1).unwrap(),
+        word: row.get(2).unwrap(),
+        kind: row.get(3).unwrap(),
+    })).unwrap();
+
+    let data = rows
+        .map(|r| r.unwrap())
+        .collect::<Vec<Abcd>>();
+
+    assert_eq!(data, vec![
+        Abcd { alpha: "A".to_string(), number: 10.0, word: "ten".to_string(), kind: "even".to_string() },
+        Abcd { alpha: "B".to_string(), number: 11.0, word: "eleven".to_string(), kind: "odd".to_string() },
+        Abcd { alpha: "C".to_string(), number: 12.0, word: "twelve".to_string(), kind: "even".to_string() },
+        Abcd { alpha: "D".to_string(), number: 13.0, word: "thirteen".to_string(), kind: "odd".to_string() },
+        Abcd { alpha: "E".to_string(), number: 14.0, word: "fourteen".to_string(), kind: "even".to_string() },
+        Abcd { alpha: "F".to_string(), number: 15.0, word: "fifteen".to_string(), kind: "odd".to_string() },
+    ]);
+}